 |
Claims  |
|
|
What is claimed is:
1. A blood pressure measurement apparatus, comprising:
blood vessel information detecting means for detecting vibration of a blood
vessel to produce a signal waveform;
storage means for storing the signal waveform produced by said blood vessel
information detecting means and time information associated with the
signal waveform;
reference maximum value point locating means for locating a reference
maximum value point of signal waveform stored in said storage means, the
reference maximum value point used as a base point for time duration
measurement;
first C-value searching means for finding a first minimum value point
within a first predetermined time duration preceding the reference maximum
value point in the signal waveform stored in said storage means;
second C-value searching means for finding an initial maximum value point
within a second predetermined time duration preceding the first minimum
value point in the signal waveform stored in said storage means;
third C-value searching means for finding a second minimum value point
within a third predetermined time duration following the reference maximum
value point in the signal waveform stored in said storage means;
discriminating means for determining whether level differences between the
second minimum value point and the reference maximum value point, the
initial maximum value point and the first minimum value point, and the
first minimum value point and the reference maximum value point each falls
within a predetermined range; and
control means for controlling said storage means, said first, second and
third C-value searching means and said discriminating means, said control
means including means for recognizing a representation of a Korotkoff
sound in the signal waveform in dependence upon the determining of said
discriminating means.
2. An apparatus according to claim 1 wherein said control means further
includes restart means for restarting with the searching for the reference
maximum value point when any of the level differences is determined to be
outside the predetermined range.
3. An apparatus according to claim 1, wherein said blood vessel information
detecting means includes setting means for setting a threshold value for
detection of the vibration of the blood vessel in accordance with the
magnitude of the representation of a Korotkoff sound recognized
immediately before.
4. An apparatus according to claim 1, wherein said storage means
periodically stores a portion of the signal waveform using a predetermined
time period.
5. A blood pressure measurement apparatus, comprising:
blood vessel information detecting means for detecting vibration of a blood
vessel to produce a signal waveform;
storage means for storing the signal waveform produced by said blood vessel
information detecting means and time information associated with the
signal waveform;
reference minimum value point locating means for locating a reference
minimum value point of the signal waveform stored in said storage means,
the reference minimum value point used as a base point for time duration
measurement;
first C-value searching means for finding a first maximum value point
within a first predetermined time duration preceding the reference minimum
value point in the signal waveform stored in said storage means;
second C-value searching means for finding an initial minimum value point
within a second predetermined time duration preceding the first maximum
value point in the signal waveform stored in said storage means;
third C-value searching means for finding a second maximum value point
within a third predetermined time duration following the reference minimum
value point in the signal waveform stored in said storage means;
discriminating means for determining whether level differences between the
second maximum value point and the reference minimum value point, the
initial minimum value point and the first maximum value point, and the
first maximum value point and the reference minimum value point each falls
within a predetermined range; and
control means for controlling said storage means, said first, second and
third C-value searching means and said discriminating means, said control
means including means for recognizing a representation of a Korotkoff
sound in the signal waveform in dependence upon the determining of said
discriminating means
6. An apparatus according to claim 5, wherein said control means further
includes restart means for restarting with the search for the reference
minimum value point when any of the level differences is determined to be
outside the predetermined range.
7. An apparatus according to claim 5, wherein said blood vessel information
detecting means includes setting means for setting a threshold value for
detection of the vibration of the blood vessel in accordance with the
magnitude of the representation of a Korotkoff sound recognized
immediately before.
8. The apparatus according to claim 5, further comprising inverting means
for inverting, with respect to a reference level, the signal waveform upon
location of the reference minimum value point by said reference minimum
value point locating means to produce an output having a value used as a
reference for maximum/minimum value detection and level determination
performed by at least said first, second and third C-value searching means
and said discriminating means.
9. The apparatus according to claim 5, wherein said storage means
periodically stores a portion of the signal waveform using a predetermined
time period.
10. A blood pressure measurement method, comprising the steps of:
(a) detecting blood vessel information to produce a signal waveform of
vibration produced by a blood vessel;
(b) storing the signal waveform produced in step (a) together with
associated time information;
(c) locating a reference maximum value point of the signal waveform stored
in step (b)
(d) searching the signal waveform stored in step (b) for a first minimum
value point within a first predetermined time duration preceding the
reference maximum value point previously located in step (c);
(e) searching the signal waveform stored in step (b) for an initial maximum
value point within a second predetermined time duration preceding the
first minimum value point previously located in step (d)
(f) searching the signal waveform stored in step (b) for a second minimum
value point within a third predetermined time duration following the
reference maximum value point previously located in step (c)
(g) determining whether level differences between the second value point
and the difference maximum value point, the initial maximum value point
and the reference maximum value point each falls within a predetermined
range;
(h) controlling execution of at least steps (c)-(g) in dependence upon said
determining in step (g); and
(i) recognizing a representation of a Korotkoff sound in the signal
waveform in dependence upon said determining in step (g).
11. A blood pressure measurement method, comprising the steps of:
(a) detecting blood vessel information to produce a signal waveform of
vibration produced by a blood vessel;
(b) storing the signal waveform proceed in step (a) together with
associated time information;
(c) locating a reference minimum value point of the signal waveform stored
in step (b);
(d) searching the signal waveform stored in step (b) for maximum value
point within first predetermined time duration preceding the reference
minimum value point previously located in step (c);
(e) searching the signal waveform stored in step (b) for an initial minimum
value point within a second predetermined time duration preceding the
first maximum value point previously located in step (d);
(f) searching the signal waveform stored in step (b) for a second maximum
value point within a third predetermined time duration following the
reference minimum value point previously located in step (c);
(g) determining whether level differences between the second maximum value
point and the reference minimum value point, the initial minimum value
point and the first maximum value point, and the first maximum value point
and the reference minimum value point each falls within a predetermined
range;
(h) controlling execution of at least steps (c)-(g) in dependence upon said
determining in step (g); and
(i) recognizing a representation of a Korotkoff sound in the signal
waveform in dependence upon said determining in step (g).
12. A blood pressure measurement method, comprising the steps of:
(a) detecting vibration of a blood vessel to produce a signal waveform;
(b) storing the signal waveform proceed in step (a) together with time
information associated therewith;
(c) locating a reference extreme value point of the signal waveform stored
in step (b) in dependence upon a threshold having a predetermined initial
value and subsequent values determined by the magnitude of the reference
extreme value point in an immediately previously recognized Korotkoff
sound;
(d) searching the signal waveform stored in step (b) for a first preceding
extreme value point within a first predetermined time duration preceding
the reference extreme value point previously located in step (c);
(e) searching the signal waveform stored in step (b) for second preceding
extreme value point within a second predetermined time duration preceding
the first preceding extreme value point previously located in step (d);
(f) searching the signal waveform stored in step (b) for a following
extreme value point within a third predetermined time duration following
the reference extreme value point previously located in step (c);
(g) determining whether level differences between the following extreme
value point and the reference extreme value point, the first preceding
extreme value point and the second extreme value point, and the first
extreme value point and the reference extreme value point each falls
within a predetermined range;
(h) controlling execution of at least steps (c)-(g) in dependence upon said
determining in step (g); and
(i) recognizing a representation of a Korotkoff sound in the signal
waveform in dependence upon said determining in step (g). |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a blood pressure measurement apparatus and method
in which a waveform discrimination method is used in the recognition of
Korotkoff sounds in the measurement of blood pressure by auscultation.
2. Description of the Prior Art
In the detection of Korotkoff sounds according to the prior art, the most
widespread approach is a discrimination method using a filter and
comparator. This is referred to as the filter comparator method. Another
approach used much less widely is a discrimination method, namely a
pattern recognition method, which is based on the waveform of the
Korotkoff sounds.
It is known that the spectral distribution of Korotkoff sounds generally
has a frequency component different from body movement and external noise.
The filter comparator method utilizes this fact and measures blood
pressure by filtering a signal detected by a microphone attached to a
pressure cuff fastened to a patient's arm, reducing the amplitude of
frequency components other than the frequency component of the Korotkoff
sounds, then comparing the frequency component of the Korotkoff sounds
with a preset threshold value by means of a voltage comparator, and
discriminating this frequency component based on its magnitude.
However, the frequency component of Korotkoff sounds not only varies from
one patient to another but also differs for one and the same patient
depending upon such measurement conditions as the time at which
measurement is made and cuff pressure Moreover, since the frequency band
of interest is fairly wide, ranging from several tens of Hertz to 200-300
Hz, it is very difficult to extract solely the Korotkoff sound component
by removing the sound of the patient's pulse and noise.
When the frequency component of the Korotkoff sounds is small in comparison
with the sound of the patient's pulse, it is difficult to distinguish
between the pulse sound and the Korotkoff sounds. Furthermore, since the
discrimination is made based on a voltage level, measurement precision is
readily influenced by any disparity in the amplitude of the Korotkoff
sounds.
The aforementioned pattern recognition method in which discrimination is
made based on the waveform of the Korotkoff sounds has recently been put
into partial practical use.
In general, the waveform of a Korotkoff sound is as shown in FIG. 2(A). The
waveform is subjected to an A-D conversion so as to make it easier to
process the sound data detected by a pick up, with the digital signal
resulting from the conversion being stored in means such as a memory. This
is referred to as pattern detection processing. Next, maximum and minimum
values are calculated from the stored signal values. For example,
characteristic points are successively detected, as shown at C1, C2, C3,
C4 in FIG. 3(A), four of such points being the minimum number necessary.
This is referred to as a characteristic point plotting step. After the
characteristic points have been detected, the general position of each
characteristic point is verified and a decision is rendered as to whether
the waveform is indeed indicative of a Korotkoff sound. This step is
referred to as a discrimination processing step. Thus, recognition
processing is divided into three process blocks.
If a characteristic point is not detected in the characteristic point
plotting processing step, the pattern detection processing step is
returned to for further signal read in. If a decision is rendered in the
discrimination step to the effect that the waveform is not that of a
Korotkoff sound, processing is executed for detecting further
characteristic points or for reading in a new signal.
The relationship among the pattern recognition processing blocks is
illustrated in FIG. 6.
A problem encountered in the pattern recognition approach is that in the
actual measurement data obtained from a living body, a fine ripple shown
in FIG. 7 tends to be produced in the vicinity of the maximum and minimum
points of the Korotkoff sound signal owing to the influence an A-D
conversion error.
Accordingly, with the method of detecting maximum and minimum values one by
one while traversing the signal waveform in regular order and then
treating each such value as a characteristic point, there is very large
amount of feedback from the discrimination processing and, hence, the
method requires a considerable period of time for execution. In addition,
there is strong possibility that characteristic points will be detected
erroneously.
SUMMARY OF THE INVENTION
Accordingly, an object of the present invention is to provide a blood
pressure measurement apparatus and method adapted to detect a maximum
value or minimum value of an extreme point capable of being traversed by a
Korotkoff sound signal constituent within a predetermined time region in
which detected characteristic points of detected blood vessel information
serve as a reference, perform a Korotkoff sound recognition by comparing
characteristic points, and recognize Korotkoff sounds accurately without
the influence of a ripple component produced in the vicinity of the
extreme values of the Korotkoff sounds.
Another object of the present invention is to provide a blood pressure
measurement apparatus and method in which a Korotkoff sound recognition is
performed with detected characteristic points serving as a reference,
whereby extreme value points having the greatest certainty of being
detected in a Korotkoff signal waveform can be treated as reference
characteristic points.
According to the present invention, the foregoing objects are attained by
providing a blood pressure measurement apparatus comprising: blood vessel
information detecting means for detecting a signal waveform of a sound or
vibration produced by a blood vessel; holding means for holding the signal
waveform detected by the blood vessel information detecting means; maximum
point detecting means for detecting a maximum point (C3) of the waveform
held by the holding means; first C-point detecting means for detecting a
minimum value point (C2) within a predetermined time region (t.sub.1) the
final instant of which is the maximum point (C3) detected by the maximum
point detecting means; first discriminating means for discriminating
whether a level difference between the minimum value point (C2) detected
by the first C-point detecting means and the maximum point (C3) falls
within a predetermined range; second C-point detecting means for detecting
a maximum value point (Cl) within a predetermined time region (t.sub.2)
the final instant of which is the detected minimum value point (C2), this
being performed when the first discriminating means discriminates that the
level difference falls within the predetermined range; second
discriminating means for discriminating whether a level difference between
the maximum value point (Cl) detected by the second C-point detecting
means and the minimum value point (C2) falls within a predetermined range;
third C-point detecting means for detecting a minimum value point (C4)
within a predetermined time region (t.sub.3) the starting instant of which
is the maximum point (C3), this being performed when the second
discriminating means discriminates that the level difference falls within
the predetermined range; third discriminating means for discriminating
whether a level difference between the minimum value point (C4) detected
by the third C-point detecting means and the maximum point (C3) falls
within a predetermined range; and a control unit for starting at least the
three C-point detecting means and three discriminating means, advancing
control successively when the respective conditions are realized, and
recognizing a Korotkoff sound in the signal waveform when the third
discriminating means discriminates that the level difference between the
minimum value point (C4) and maximum point (C3) falls within the
predetermined range.
According to a preferred embodiment of the present invention, the control
unit includes control means for restarting control from detection of the
maximum point (C3) when the discrimination condition for any of the level
discriminating means fails to be satisfied, and the blood vessel sound
detecting means includes setting means for setting a threshold value of a
detection signal in accordance with the magnitude of a Korotkoff sound
recognized immediately before.
Further, the control unit includes holding means for holding the signal
waveform at every predetermined time.
The foregoing objects may also be attained by providing a blood pressure
measurement apparatus comprising: blood vessel information detecting means
for detecting a signal waveform of a sound or vibration produced by a
blood vessel; holding means for holding the signal waveform detected by
the blood vessel information detecting means; minimum point detecting
means for detecting a minimum point (C3) of the waveform held by the
holding means; first C-point detecting means for detecting a maximum value
point (C2) within a predetermined time region (t.sub.1) the final instant
of which is the minimum point (C3) detected by the minimum point detecting
means; first discriminating means for discriminating whether a level
difference between the maximum value point (C2) detected by the first
C-point detecting means and the minimum point (C3) falls within a
predetermined range; second C-point detecting means for detecting a
minimum value point (Cl) within a predetermined time region (t.sub.2) the
final instant of which is the detected maximum value point (C2), this
being performed when the first discriminating means discriminates that the
level difference falls within the predetermined range; second
discriminating means for discriminating whether a level difference between
the minimum value point (Cl) detected by the second C-point detecting
means and the maximum value point (C2) falls within a predetermined range;
third C-point detecting means for detecting a maximum value point (C4)
within a predetermined time region (t.sub.3) the starting instant of which
is the minimum point (C3), this being performed when the second
discriminating means discriminates that the level difference falls within
the predetermined range; third discriminating means for discriminating
whether a level difference between the maximum value point (C4) detected
by the third C-point detecting means and the minimum point (C3) falls
within a predetermined range; and a control unit for starting at least the
three C-point detecting means and three discriminating means, advancing
control successively when the respective conditions are realized, and
recognizing a Korotkoff sound in the signal waveform when the third
discriminating means discriminates that the level difference between the
maximum value point (C4) and minimum point (C3) falls within the
predetermined range.
The control unit includes control means for restarting control from
detection of the minimum point (C3) when the discrimination condition for
any of the level discriminating means fails to be satisfied.
The blood vessel sound detecting means includes setting means for setting a
threshold value of a detection signal in accordance with the magnitude of
a Korotkoff sound recognized immediately before.
The apparatus further includes inverting means for inverting, with respect
to a reference level, a signal waveform held by detection of the minimum
point by the minimum point detecting means, wherein a value of an output
inverted by the inverting means is used as a reference for maximum/minimum
value detection and level discrimination performed by at least the three
C-point detecting means and three discriminating means.
The control unit includes holding means for holding the signal waveform at
every predetermined time.
Further, according to the present invention, there is provided a blood
pressure measurement method which comprises: a blood vessel information
detecting step for detecting a signal waveform of a sound or vibration
produced by a blood vessel; a holding step for holding the signal waveform
detected at the blood vessel information detecting step; a maximum point
detecting step for detecting a maximum point (C3) of the waveform held at
the holding step; a first C-point detecting step for detecting a minimum
value point (C2) within a predetermined time region (t.sub.1) the final
instant of which is the maximum point (C3) detected at the maximum point
detecting step; a first discriminating step for discriminating whether a
level difference between the minimum value point (C2) detected at the
first C-point detecting step and the maximum point (C3) falls within a
predetermined range; second C-point detecting means for detecting a
maximum value point (Cl) within a predetermined time region (t.sub.2) the
final instant of which is the detected minimum value point (C2), this
being performed when it is discriminated at the first discriminating step
that the level difference falls within the predetermined range; a second
discriminating step for discriminating whether a level difference between
the maximum value point (Cl) detected at the second C-point detecting step
and the minimum value point (C2) falls within a predetermined range; a
third C-point detecting step for detecting a minimum value point (C4)
within a predetermined time region (t.sub.3) the starting instant of which
is the maximum point (C3), this being performed when it is discriminated
at the second discriminating step that the level difference falls within
the predetermined range; a third discriminating step for discriminating
whether a level difference between the minimum value point (C4) detected
at the third C-point detecting step and the maximum point (C3) falls
within a predetermined range; and a step of starting at least the three
C-point detecting steps and three discriminating steps, advancing control
successively when the respective conditions are realized, and recognizing
a Korotkoff sound in the signal waveform when it is discriminated at the
third discriminating step that the level difference between the minimum
value point (C4) and maximum point (C3) falls within the predetermined
range.
The present invention further provides a blood pressure measurement method
which comprises: a blood vessel information detecting step for detecting a
signal waveform of a sound or vibration produced by a blood vessel; a
holding step for holding the signal waveform detected at the blood vessel
information detecting step; a minimum point detecting step for detecting a
minimum point (C3) of the waveform held at the holding step; a first
C-point detecting step for detecting a maximum value point (C2) within a
predetermined time region (t.sub.1) the final instant of which is the
minimum point (C3) detected at the minimum point detecting step; a first
discriminating step for discriminating whether a level difference between
the maximum value point (C2) detected at the first C-point detecting step
and the minimum point (C3) falls within a predetermined range; second
C-point detecting means for detecting a minimum value point (Cl) within a
predetermined time region (t.sub.2) the final instant of which is the
detected maximum value point (C2), this being performed when it is
discriminated at the first discriminating step that the level difference
falls within the predetermined range; a second discriminating step for
discriminating whether a level difference between the minimum value point
(Cl) detected at the second C-point detecting step and the maximum value
point (C2) falls within a predetermined range; a third C-point detecting
step for detecting a maximum value point (C4) within a predetermined time
region (t.sub.3) the starting instant of which is the minimum point (C3),
this being performed when it is discriminated at the second discriminating
step that the level difference falls within the predetermined range; a
third discriminating step for discriminating whether a level difference
between the maximum value point (C4) detected at the third C-point
detecting step and the minimum point (C3) falls within a predetermined
range; and a step of starting at least the three C-point detecting steps
and three discriminating steps, advancing control successively when the
respective conditions are realized, and recognizing a Korotkoff sound in
the signal waveform when it is discriminated at the third discriminating
step that the level difference between the maximum value point (C4) and
minimum point (C3) falls within the predetermined range.
Other features and advantages of the present invention will be apparent
from the following description taken in conjunction with the accompanying
drawings, in which like reference characters designate the same or similar
parts throughout the figures thereof.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram illustrating the basic construction of a
Korotkoff sound recognition apparatus embodying the present invention;
FIGS. 2(A) and 2(B) are views showing typical patterns of Korotkoff sounds;
FIGS. 3(A) and 3(B) are views showing characteristic portions of a
Korotkoff sound waveform;
FIGS. 4(A) and 4(B) is a flowchart illustrating processing extending from
detection of each characteristic point of a Korotkoff sound to recognition
of a Korotkoff sound according to an embodiment of the present invention;
FIGS. 5(A) to 5(O) are views illustrating the recognition states of each
characteristic point of a Korotkoff sound waveform when the processing
indicated by the flowchart of FIG. 4 is executed;
FIG. 6 is a block diagram illustrating a conventional Korotkoff sound
discrimination method based on waveform configuration; and
FIG. 7 is a view showing a Korotkoff signal waveform in which a small
ripple is produced in the vicinity of extreme values.
DESCRIPTION OF THE PREFERRED EMBODIMENT
An embodiment of the present invention will now be described in detail with
reference to the drawings.
FIG. 1 is a block diagram illustrating the basic construction of an
embodiment of the present invention. The arrangement includes a microphone
1 for picking up Korotkoff sounds (hereafter referred to as "K-sounds")
and for producing an analog output signal indicative thereof. The input
analog signal between range of 0.3V and 2.0V is converted into a 8 bit
digital signal every 4 milliseconds (ms) by an analog-digital (A/D)
converter 2 before being applied to an arithmetic circuit 3. The latter
serves as recognition means and is adapted to recognize a K-sound by
processing a series of sound data signals obtained in digital form from
the A/D converter 2. The arithmetic circuit 3 comprises a one-chip CPU
having a RAM and a ROM and is so illustrated that the various functions
implemented by executing a program stored in the ROM are shown in block
form. The present invention is capable of implementing these functions
efficiently with the limited memory and limited processing time given the
CPU. Numeral 4 denotes a display unit for displaying the fact that a
K-sound has been recognized, for displaying other information as well.
The arithmetic circuit 3 includes a data read-in unit 5 for reading the
digital output signal from the A/D converter 2 into the arithmetic circuit
3, a threshold value setting unit 6 which compares the newly read digital
signal data from the data read-in unit 5 and the threshold value
determined based upon the most recent K-sound. The arithmetic circuit 3
further includes a time generator for generating time information, and a
memory (RAM) 9 for storing a digital data as well as the time information
prevailing at the instant of detection. The threshold value which is to be
held in the threshold value setting unit 6 is calculated from the most
recent K-sound in a K-sound recognition unit 15 in accordance with the
following equation,
.vertline.C3-P.sub.o .vertline..multidot..beta.l, where: . . . (1)
C3: the characteristic point of the detected K-sound
P.sub.o : reference level
.beta.l: one-third (1/3)
If the newly read digital signal is larger than the most recent threshold
value, the sign bit is set to "1". This sign bit is included in the output
106. If, on the contrary, the new data is smaller than the threshold
value, the sign bit is set to "0". After completion of this operation, the
digital data signal containing the sign bit is delivered to the memory 9
as digital signal data 106. The sign bit is referred to by a C3 detector
10 in order to determine whether to produce a signal on line 113 or not.
Before continuing with the description of the functional blocks of the
apparatus shown in FIG. 1, let us discuss typical patterns of the K-sounds
which are to be recognized by the apparatus.
FIG. 2(A) is a typical pattern of a K-sound waveform recognized by the
apparatus of the illustrated embodiment, and FIG. 2(B) illustrates the
pattern when the signal level is inverted. As it is possible that the
input waveform may represent two opposite waveforms, drawings illustrating
the respective waveforms, for example, FIGS. 2(A) and 2(B), are provided
in this application. The characteristic points of the K-sound waveform are
the four points C1-C4 shown in FIGS. 3(A), (B). In order to help
understanding of the present invention, level differences dP1 to dP3 and
time regions T.sub.1 to T.sub.3 are labeled in FIG. 3A. They will be
referred to in the following discussion. In the illustrated embodiment, a
K-sound is recognized on the basis of the relationship among these four
points.
In FIGS. 3(A), (B), the point C3 is defined as the point where the signal
level attains an absolute extreme valve, i.e., the highest peak or lowest
valley, and is a portion which has great significance for the purpose of
K-sound recognition, described below. Specifically, once the
characteristic point C3 has been found, the characteristic points C1, C2,
C4 are each obtained by a prescribed analytical method which starts from
the point C3.
Each of the functional blocks described below constitutes means for
recognizing the abovementioned K-sound waveform patterns both reliably and
efficiently.
Returning to FIG. 1, numeral 10 is the C3 detector for detecting a relative
(local) extreme valve, i.e., a maximum value or minimum value in the
digital signal data read out of the memory 9. Numeral 11 designates a
level inverter which, for the purpose of K-sound recognition, inverts the
level of the signal waveform data, which is read out of the memory 9,
whenever necessary. A characteristic point detector 12 performs a
predetermined calculation with regard to the signal waveform data read out
of the memory 9 and time data to check for the presence of a waveform
located at each of the characteristic points C2, C2, C4. The
characteristic point detector 12 comprises a time region setting unit 13
for generating prescribed time region data, and a K-sound discriminator 14
for discriminating whether sound data of a signal level forming a
characteristic is present within the time region, and is adapted to detect
each characteristic point in accordance with a predetermined calculation
procedure, described below, when a signal indicating that the C3 point has
been detected is received from the C3 detector 10. The output of the
K-sound discriminator 14 is applied to a K-sound recognition unit 15,
which examines the positional relationship among a collection of
characteristic points found by the characteristic point detector 12, in
order to recognize a K-sound.
The operation of the present embodiment comprising the foregoing elements
will now be described.
The output of the microphone 1 is an analog electric signal 101 indicative
of a K-sound picked up by the microphone. The signal 101 is converted into
a digital signal 102 at every 4ms sampling instant by the A/D converter 2.
The digital signal 102 at the output of the A/D converter 2 is read into
the arithmetic circuit 3 by the data read-in unit 5 and is applied to the
threshold value setting unit 6 as a series of digital signal data 103 in a
time series. The threshold value setting unit 6 sets a threshold value in
dependence upon a signal 117 from the K-sound recognition unit 15
indicative of the magnitude of a K-sound which appeared last in accordance
with equation 1. By setting the threshold value to .vertline.C3-P.sub.o
.vertline..multidot..beta.l, the unit 6 reduces the influence of noise
contained in the waveform data 103.
In order to suitably deal with a signal pattern input of any amplitude
whatsoever at the start of measurement, no threshold value is set when
measurement starts After measurement starts, however, a threshold value is
set upon predicting, from the magnitude of an immediately preceding
K-sound, the smallest magnitude capable of being traversed by the next
K-sound. More specifically, in a case where a K-sound has already appeared
in the course of measurement, the threshold value setting unit 6 sets a
threshold value in dynamic fashion in dependence upon the signal 117 from
the K-sound discriminator 15 indicative of the magnitude of the threshold
.vertline.C3-P.sub.o .vertline..multidot..beta.l, thereby making it
possible to detect the C3 point accurately and rapidly.
Accordingly, the threshold value setting unit 6 in the apparatus of the
illustrated embodiment is different in nature from threshold value setting
means in the conventional comparator method, in which a threshold value is
set that is fixed with respect to the amplitude of the K-sound. The
threshold value setting unit 6 delivers the digital signal data 106 to the
memory 9 each 4 ms interval and also delivers timing signal 105 to the
time generator 8 each sampling instant at the A/D converter 2.
The time generator 8 is a unit which cyclically counts timing information
that increases every millisecond, by way of example. When the timing
signal 105 is received from the threshold value setting unit 6, the time
generator 8 successively counts up a write-in address 120 of the memory 9
so that the digital signal data 106 from the threshold value setting unit
6 and prevailing clocked time information 107 are written into the memory
in accordance with the counted up address. Thus, the digital signal value
data 106 and the time information 107 prevailing at the moment of
detection are stored in the memory 9.
The time generator 8 also outputs a read-out address 120 of the memory 9 at
a predetermined time interval and produces a read enable signal 121 when a
read-out becomes possible.
The digital signal data 106 stored in memory 9 is read by the C3 detector
10 and characteristic point detector 12 in accordance with the read enable
signal 121, whereby K-sound recognition processing is performed. In order
that the memory 9 can be read from any address when K-sound recognition
processing is executed, the characteristic point detector 12 provides the
time generator 8 with an address designating signal 122 for designating a
read-out address from which a read-out is to be started.
The K-sound recognition processing will now be described with reference to
the flowchart of FIGS. 4(A) and 4(B).
In accordance with the read enable signal 121 from the time generator 8,
the C3 detector 10 reads sound data 108 out of the memory 9 in accordance
with the successively stored time series, examines these data in regular
order and executes processing for detecting the C3 point in the signal
pattern shown in FIG. 3(A) or FIG. 3(B).
In the first step S90 of the flowchart shown in FIGS. 4(A) and (B), an
inversion flag 10a internally of the C3 detector 10 is set to "0". When
the inversion flag 10a is "0", an inversion indicating signal 109 is
reset; when the flag 10a is "1", the inversion designating flag 109 is
set. When the inversion indicating signal 109 is in the reset state, the
level inverter 11 delivers read-out data 110 from memory 9 directly to the
characteristic point detector 12 as output data 111. When the inversion
indicating signal 109 is in the set state, the level inverter 11 inverts
the read-out data from the memory 9 and delivers the result to the
characteristic point detector 12 as the output data 111.
Initially, the inversion flag 10a is set to "0" and the inversion
designating signal 109 is reset. Consequently, the read-out data from
memory 9 is applied as such to the characteristic point detector 12.
Next, at a step S91, the C3 detector 10, in accordance with the read enable
signal 121, reads the digital signal data 106 from the threshold value
setting unit 6 stored successively in memory 9 via, the A/D converter 2
out of the memory in the same order in which it was stored and compares
this with digital signal data 106 read out immediately before. The time
generator 8 exercises read-out control separate from the write-in of the
digital signal data 106. The reading of data from the memory 9 can be
performed immediately by writing in the digital signal data 106 by means
of the threshold value setting unit 6. Extreme value detection processing
is executed from step S92 onward and is performed by a level comparison of
digital signals at three consecutive points in the sound data 108.
Thus, at the step S92, digital signals at three consecutive points are
compared to determine whether a valley point has been detected, that is,
to check whether the level difference between adjacent ones of the points
changes from a decreasing value to an increasing value. If a valley pont
is detected and this valley point is given a sign bit "1", that is the
newly obtained digital signal data is determined by the C3 detector 10 to
be larger than the current threshold value, this point is treated as being
the characteristic point C3 and the program proceeds from the step S92 to
a step S94, at which the inversion flag 10a is set to "1" and the
inversion indicating signal 109 is delivered to the level inverter 11. The
program then proceeds to a step S95.
When the inversion indicating signal 109 is delivered to the level inverter
11, K-sound detection is performed. To this end, the level of each item of
waveform data 110 corresponding to the characteristic points C1-C4 in FIG.
3(B) and read out of the memory 9 is inverted with regard to a base line
[level P.sub.o shown in FIGS. 5(D), (E)] that will turn these levels into
the signal pattern shown in FIG. 3(A). The inverted levels are delivered
to the characteristic point detector 12.
If a valley point is not detected at the step S92, the program proceeds to
a step S93, at which it is determined whether a peak C3 has been detected,
that is, whether the level difference between adjacent ones of the three
consecutive points changes from an increasing value to a decreasing value.
If a peak is detected, this peak is determined as to whether the peak is
exceeding the threshold value with referring to the sign bit included in
the digital signal data 106. If the peak is exceeding the threshold value,
the peak is treated as being the characteristic point C3 and the program
proceeds to a step S95. If a peak is not detected at the step S93, the
program returns to the step S91, the next item of digital signal data is
read and processing for detecting a characteristic point C3 is performed
again.
If a peak point is detected at the step S93, the inversion flag 10a remains
at "0" and the program proceeds to the step S95. This step calls for a
characteristic point detection signal 113 to be sent from the C-point
detector 10 to the characteristic point detector 12 to indicate that the
detected point is a characteristic point. The program then proceeds to a
step S104.
If a K-sound is recognized, what is detected first is the peak point. An
example of the state in which the initial peak is detected is illustrated
in FIG. 5(A).
In response to the characteristic point detection signal 113 from the C3
detector 10 indicating that the C3 point has been detected, the
characteristic point detector 12 initiates detection of each
characteristic point of the digital signal data constituting a K-sound.
This being performed by processing from step S104 onward.
When the characteristic point detection signal 113 is received, the
characteristic point detection circuit 12 produces the address designating
signal 122 so that data stored in the memory 9 prior to detection of the
characteristic point are read out of the memory sequentially in the same
order that the data were stored. These data are stored in a RAM. In other
words, each item of data from C1 to C3 is stored every 4 ms sampling
period in the RAM at the instant C3 is detected.
The step S104 calls for the time region setting unit 13 to set a
predetermined time region t.sub.1 the final instant of which is the
position of C3. The unit 13 produces a time region signal 114 indicative
of this time region and applies the signal to the K-sound discriminator
14. The setting of this time region can be accomplished by storing in a
ROM a predetermined value in accordance with the figures given in below.
The set time region t.sub.1 is illustrated in FIG. 5(B). According to the
embodiment of the present invention, each time region t.sub.1, t.sub.2 and
t.sub.3 have the values t.sub.1 =10, t.sub.2 =15 and t.sub.3 =15
[.times.4ms]. And these time region data have been stored previously in
the ROM which is constituting the time region setting unit 13.
Next, the program proceeds to a step S105, at which the K-sound
discriminator 14 reads digital signal data within the time region t.sub.1
set by the time region setting unit 13 and stored in the RAM, detects a
minimum value within the read data and treats this value as C2 [FIG.
5(C)]. The minimum level point is detected by comparing the levels of two
points in the output data 111 from the level inverter 11. Next, a step
S106 calls for a decision as to whether the level difference (dP2) between
C2 and C3 falls within a predetermined range. The upper and lower limits
of this range are stored beforehand in the ROM in accordance with the
below table. According to the embodiment of the present invention, each
level difference dP1, dP2 and dP3 is given as follows.
______________________________________
lower limit
upper limit
______________________________________
dP2 15 --
dP1 dP2 .times. .alpha.1
dP2 .times. .alpha.2
dP3 dP1 .times. .alpha.3
dP2 .times. .alpha.4
______________________________________
In the above table, the unit is 0.7V/256 and also, the level differences
illustrated in FIG. 3A, dP1, dP2 and dP3, are given below.
dP1: voltage difference between C1 and C2
dP2: voltage dif | | |
