 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a method of identifying an object by detecting a
contour line of the object to be detected by using an image pick up
apparatus, for example, a television camera or the like.
2. Description of the Prior Art
A convertional method of identifying an article or object to be identified
utilizes the edges of respective surfaces of the objects, and according to
this method, the point of the picture image at which the brightness
changes abruptly are extracted as the edges, and the edges are connected
together to convert them into a line picture.
The steps of identifying a circular body with the contour line extraction
method will be described with reference to FIGS. 1a through 1d. At first,
an original picture image (shown in FIG. 1a) photographed with a
television camera is differentiated along respective scanning lines to
extract a contour candidate point at which the brightness changes rapidly
(see FIG. 1b). Then, respective picture elements near this point are
differentiated. Among the differentiated values, a picture element having
the maximum differentiated value is taken as a point continuous to the
contour candidate point. This processing is repeated a certain number of
times to obtain continuous contour points (contour line candidate point)
(see FIG. 1c) and when these candidate points are closed (see FIG. 1d),
they are deemed as an object.
According to this prior art contour extraction method, however, the tracing
of the contour candidate points are rendered difficult by the following
factors:
(1) blooming caused by metal luster (see FIG. 2a),
(2) overlapping of objects (see FIG. 2b),
(3) vague or not clear image caused by rust, spoil, etc. of the object
surfaces, and
(4) distortion of the picture image caused by electrical noise.
As a consequence, there is a defect that an actually presenting object
would not be detected. Furthermore, identification algorithm for solving
these problems becomes complicated so that real time processing is almost
impossible.
SUMMARY OF THE INVENTION
It is an object of this invention to provide a novel method of identifying
an object capable of simplifying the identification algorithm of an object
to be detected, shortening the time of processing a picture image and
increasing the percentage of correctly detecting the object.
According to this invention, there is provided a method of identifying an
object of the type wherein an object to be detected is photographed, and
the object is identified by detecting a portion of a contour line of a
photographed picture image, characterized in that the method comprises the
steps of preparing filter means; scanning the picture image with the
filter means for designating a first group of a predetermined number of
continuous picture elements and a second group of a plurality of picture
elements spaced from the first group by a distance corresponding to a
contour line of the object; detecting presence or absence of a contour
candidate point at which brightness changes rapidly in respective picture
element groups; and identifying that the object presents when the contour
candidate points are simultaneously detected in respective picture element
groups.
According to a modification of this invention, there is provided a method
of identifying an object wherein a portion of a contour line of the object
to be detected comprises parallel lines, characterized in that the method
comprises the steps of: preparing filter means; designating with the
filter means a pair of picture element groups such that the
center-to-center distance thereof will correspond to a spacing between the
parallel lines; recognizing a scanning position at which contour candidate
points are simultaneously detected in the picture element groups as a
filter characteristic point; after detecting the filter characteristic
point, trace scanning in an auxiliary scanning direction from the scanning
position for detecting presence or absence of the filter characteristic
point; and identifying presence of the object to be detected when a traced
length becomes larger than a preset length.
According to still another embodiment of this invention, there is provided
a method of identifying an object having substantially circular contour
line, characterized by the steps of preparing filter means; designating
with the filter means a pair of picture element groups having a
center-to-center distance corresponding to the diameter of the
substantially circular contour line; denoting a scanning position at which
contour candidate points are simultaneously detected in the picture
element groups as a filter characteristic point; after detecting the
filter characteristic point, trace scanning in an auxiliary scanning
direction from the scanning position for detecting presence or absence of
the filter characteristic point; determining a center position candidate
point of the substantially circular contour line from the central position
between picture element groups by the filter means and a center position
of the traced length when the traced length exceeds a preset length;
detecting presence or absence of the filter characteristic point or a
contour candidate point by sequentially rotating the filter means about
the candidate point of the center position, and then identifying presence
or absence of the object when percentage of a presence of the filter
characteristic point or the contour candidate point exceeds a preset value
.
BRIEF DESCRIPTION OF THE DRAWINGS
In the accompanying drawings:
FIGS. 1a through 1d are schematic representations showing the steps of
identifying an object according to a prior art contour line extraction
method;
FIGS. 2a and 2b show examples of the factors that render difficult to trace
a contour line according to a prior art method;
FIGS. 3a and 3b are schematic representations for explaining the principle
of this invention;
FIGS. 4 and 5 show filters utilized in this invention;
FIG. 6 is a block diagram showing one example of a logic circuit utilized
to carry out the method of this invention;
FIGS. 7, 19 and 25 show the brightness of the picture image data stored in
the RAM array shown in FIG. 6;
FIGS. 8, 16, 17, 18, 23 and 24 are flow charts showing the steps of
processings with a central processing unit (CPU) shown in FIG. 6;
FIGS. 9 and 10 are block diagrams showing different examples of the
differentiating circuit;
FIGS. 11 through 15 show other examples of the filter;
FIG. 20 is an enlarged view of a portion of FIG. 19;
FIGS. 21a, 21b and 21c and FIGS. 22a and 22b are diagrams useful to explain
how to identify parallel lines of different type; and
FIGS. 26 and 27 are diagrams useful to explain the flow chart shown in FIG.
24.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
The invention will be described hereunder with reference to the preferred
embodiments.
The principle of the method of this invention will firstly be described
with reference to FIGS. 3a and 3b. Suppose now that a dark object 1 to be
detected presents in a bright background. The brightness (black and white)
of the object 1 along a sectional line a is shown by a graph shown in FIG.
3b. Points A and B shown in FIG. 3b at which the brightness changes
rapidly are characteristic points of the object. At point A the brightness
decreases or builds down, whereas at point B the brightness increases or
builds up.
According to the method of this invention, the object is identified by
checking as to whether there are contour candidate points at which the
brightness changes rapidly at the same time in a plurality of ranges set
in connection with the contour line of the object to be detected.
Furthermore, the identification of the object is made to be more accurate
by checking whether the direction of variation in the brightness at each
candidate point coincides with a direction determined by the background
and the brightness thereof.
A filter used for identifying an object according to the present invention
will be described hereunder.
According to this invention, a filter is defined as a means for designating
the picture data having a specific locational relation among picture data
in one filed of a television camera. In other words, the term filter does
not mean optical or physical filter but is derived from its function.
More particularly, as shown in FIG. 4, the filter in this invention
designates several specific ranges A.sub.l, A.sub.2,--A.sub.n each having
specific length among picture data in one field. Each range A.sub.l,
A.sub.2,--A.sub.n consists of a group of aligned picture elements arranged
in one line having the length of .DELTA.L.sub.1,.DELTA.L.sub.2 --,
.DELTA.L.sub.n, respectively, and is spaced from the reference position P
by a distance of L.sub.l, L.sub.2, --L.sub.n, respectively.
Assuming that N represents a maximum number of contour points of an object
encountered when the object is cut transversely, the number n is chosen
from two to N (2.ltoreq.n.ltoreq.N). Since large n results in long picture
image processing time and small n results in high probability of erroneous
recognition, suitable value has to be chosen within the above-mentioned
scope (2.ltoreq.n.ltoreq.N). Spacings L.sub.1, L.sub.2,--, L.sub.n are
determined principally by the geometrical configuration of the object to
be detected and the magnifying power of the television camera,
Furthermore, the lengths of ranges .DELTA.L.sub.1,
.DELTA.L.sub.2,--L.sub.n are determined considering types of the contour
of the object corresponding to each range (3-dimensional shape), noise,
blooming, distortion of the optical system, etc.
Filters thus defined are set in advance corresponding to the contour line
of the object to be detected, an input picture is scanned by using the
filters and a check is made as to whether, there are contour candidate
points in every ranges A.sub.1, A.sub.2--, A.sub.n. (the reference
position P when there are contour candidate points in every ranges is
referred to as a filter characteristic point.)
For example, as shown in FIG. 5, a filter comprising two ranges A.sub.1 and
A.sub.2 (the spacing between the centers of these two ranges is l) is set
for an object 1 shown in FIG. 3. Then, while moving the filter position
(represented by the reference position P shown by a hatched circle), a
check is made as to whether contour candidate points concurrently present
in the ranges A.sub.l and A.sub.2.
To illustrate in a more concrete manner, assuming that one field consists
of n x m picture elements, the distance between the contour lines of the
object 1 corresponds to eleven picture elements Assuming that the lengths
.DELTA.L.sub.1 and .DELTA.L.sub.2 consists of three picture elements
respsectively and the reference position is located at a picture element e
(i, j), this filter designates six picture elements of [A.sub.1
.vertline.e(i-6, j), e(i-5, j), e(i-4, j)] and [A.sub.2.vertline. e(i+4,
j), e(i+5, j), e(i+6, j)]. It is obvious that designated picture elements
move as the filter position moves. Then a check is made as to whether
there is a filter characteristic point at which contour candidate points
exist concurrently in the range [A.sub.1 .vertline.e(i-6, j), e(i-5, j),
e(i-4,j)]and the range [A.sub.2 .vertline.e(i+4, j), e(j+5, j), e(i+6,
j)]. This check can be made by monitoring brightness difference
corresponding to each end of the ranges A.sub. 1 and A.sub.2. Alternately,
whether the directions of brightness variation in respective ranges
A.sub.1, A.sub.2, --, A.sub.n coincide with those determined by the object
and its background amy be adopted as a criterion for checking the presence
of a characteristic point. In other words, as shown in FIG. 5, when there
concurrently exist contour candidate points whose brightness varies from
light to dark in the range A.sub.1 and from dark to light in the range
A.sub.2, a filter characteristic point is judge to exist.
An industrial television (ITV) camera 2 shown in FIG. 6 photographs the
object 1 to be detected in a predetermined field of view to send a video
composite signal containing the brightness signal of the input picture
image to a synchronous isolating circuit 3 and an A/D converter 4. The
synchronous isolating circuit 3 operates to separate a synchronizing
signal from the video composite signal. The synchronizing signal thus
separated is used to designate an address of a random access memory array
(RAM array) 5, while the A/D converter 4 converts the inputted video
composite signal into a picture imag data having 16 tones of brightness
for writing the picture image data in the designated address. In this
manner, picture image data of one picture representing the brightness of
the original picture image shown in FIG. 7 are stored in the RAM array 5.
Any picture image data can be read out by designating X and Y addresses of
the RAM array 5.
A memory circuit 6 stores parameters constituting the filter as well as the
main program, (ROM) for carrying out the method of this invention, and a
central processing unit (CPU)7 executes the processings of the picture
image data stored in the RAM array 5 in accordance with the content of the
filter and the content of the main program.
Now explanation will be made as to filter parameters to be registered in
the memory 6 and its storage manner.
Various types of data sets having respective ranges and designating the
aforementioned filter (which data sets will be defined as the filter
element data sets), as will be clear from the foregoing explanation,
indicate ranges having specific intervals respectively. That is, each of
the data sets indicative of the respective ranges comprises data (first
data) indicative of the respective amounts of offset in two points
corresponding to both ends of the range from said reference position P, a
data (a second data) indicative of a threshold value as to a brightness
difference between picture elements (picture image data) corresponding to
these two points, and a data (a third data) indicative of a condition as
to a brightness gradient between the picture elements (picture image data)
corresponding to the two points.
In the case of the filter shown in FIG. 4, for example, the filter element
data set corresponding to the range A.sub.1 comprises an offset amount
data (X.sub.1,Y.sub.1) of a left end point of the range A.sub.1 from the
reference position P, an offset amount data (X'.sub.1, Y'.sub.1) of a
right end point of the range A.sub.1 from the reference point P, a
threshold value data T.sub.1 on a brightness difference between these left
and right end points, and a condition data g.sub.1 on a brightness
gradient between the left and right end points; the filter element data
set corresponding to the range A.sub.2 comprises an offset amount data
(X.sub.2,Y.sub.2) of a left end point of the range A.sub.2 from the
reference position P, an offset amount data (X'.sub.2, Y'.sub.2) of a
right end point of the range A.sub.2 from the reference point P, a
threshold value data T.sub.2 on a brightness difference between these left
and right end points, and a condition data g.sub.2 on a brightness
gradient between the left and right end points; and the filter element
data set corresponding to the range A.sub.n comprises an offset amount
data (X.sub.n, Y.sub.n) of a left end point of the range A.sub.n from the
reference position P, an offset data (X'.sub.n, Y'.sub.n) of right end
point of the range A.sub.n from the reference point P, a threshold value
data T.sub.n on a brightness difference between these left and right end
points, and a condition data g.sub.n on a brightness gradient between the
left and right end points. And these filter element data sets are
colectively registered in the aforementioned memory 6 as filter data in
such a manner as to be given in Table 1 below.
TABLE 1
______________________________________
Second
Data
First Data (Brightness
Offset amount
Offset amount
difference
Third Data
of left end
of right end
threshold
(Brightness
Contents
point from P
point from P
value) gradient)
______________________________________
Range
A.sub.1
X.sub.1, Y.sub.1
X'.sub.1, Y'.sub.1
T.sub.1 g.sub.1
A.sub.2
X.sub.2, Y.sub.2
X'.sub.2, Y'.sub.2
T.sub.2 g.sub.2
A.sub.n
X.sub.n, Y.sub.n
X'.sub.n, Y'.sub.n
T.sub.n g.sub.n
______________________________________
In the case of the filter shown in FIG. 4, since the ranges A.sub.1,
A.sub.2, and A.sub.n are all set one-dimensionally in the x-axis direction
with respect to the reference position P, the offset amounts Y.sub.1,
Y'.sub.1, Y.sub.2, Y'.sub.2, Y.sub.n and Y'n given in Table 1 above become
all "0".
In the above table, the brightness difference threshold value T as the
second data is determined as a limit value of a brightness difference
between picture elements (picture image data) corresponding to the left
and right end points so that the contour part of the corresponding
detection object can be correctly identified as a contour part on the
basis of the type (three-dimensional configuration) of the object in
connection with the width (a distance between the left and right end
points) of the corresponding range.
In the above table, further, the brightness gradient g as the third data is
set, for example, to be "+1" under a condition that the brightness of one
of the picture elements (picture image data) corresponding to right one of
the above two points (left and right end points) is higher than that of
the other picture element corresponding to the left end point, and set
conversely to be "-1" under a condition that the brightness of the picture
image data corresponding to the right end point is lower than that of the
picture image data corresponding to the left end point.
With such a filter as shown in FIG. 4, the values given in the drawing such
as L.sub.1, .DELTA.L.sub.1, have relationships with the above filter data,
which are collectively listed below.
X.sub.1 =-L.sub.1 .DELTA.L.sub.1 /2, X'.sub.1 =-L.sub.1 +.DELTA.L.sub.1 /2
X.sub.2 =-L.sub.2 -.DELTA.L.sub.2 /2, X'.sub.2 =-L.sub.2 +.DELTA.L.sub.2 /2
X.sub.n =L.sub.n -.DELTA.L.sub.n /2, X'.sub.n =L.sub.n +.DELTA.L.sub.n /2
Y'=Y'.sub.1 =Y.sub.2 =Y'.sub.2 =Y.sub.n =Y'.sub.n =0
T.sub.1, T.sub.2 and T.sub.n are arbitrarily set according to the type
(three dimensional configuration) of the corresponding contour part in
connection with }(X.sub.1, O), (X'.sub.1, O)}, {(X.sub.2, O), (X'.sub.2,
O)} and {(X.sub.n, O), (X'.sub.n, O)}.
g.sub.1 =+1, g.sub.2 =-1, and g.sub.n =-1.
The steps of processings executed by CPU 7 will be described with reference
to the flow chart shown in FIG. 8.
The CPU 7 basically functions to scan input picture image data (data stored
in the RAM array 5) on the basis of comparison with aformentioned filter
data, to detect whether or not two conditions are simultaneously
satisfied, that is, one condition that the brightness difference between
the picture image data corresponding to two points and represented by the
first data simultaneously exceeds the corresponding threshold value
represented by the second data with respect to all the filter element data
sets, and the other condition that the brightness gradient between the
picture image data corresponding to the two points satisfies the
corresponding condition given by the third data with respect to all the
filter element data sets. And when these conditions are met, the CPU 7
that the aforementioned filter characteristic point is present. More
specifically, at step 100, the filter position i.e., a reference position
P (which will be a filter supporting point for scanning) of the filter
shown in FIG. 5 is moved to the starting position P.sub.ST shown in FIG.
7, and at step 101, a check is made as to whether the scanning position is
the filter characteristic point or not.
The judgment as to whether the scanning position is the characteristic
point of the filter or not is effected by differentiating the picture
image data of two picture element groups designated by the filter.
To differentiate the picture image data after the D/A conversion, a
difference computation is effected. The following equations are generally
used for the difference computation.
Primary difference: .DELTA.f(i)=f(i)-f(i-l) (1)
Secondary difference: .DELTA..sup.2 f(i-1)=f(i)+f(i-2)-2f(i-l) 10 (2)
A differentiating circuit 8 shown in FIG. 9 is given an example of a
circuit for performing the difference computation represented by equation
(1) actually executed by the CPU 7. In the difference computation, picture
image data f(i) sequentially derived out from continuous picture element
groups designated by the filter are applied to a register 8a and a
subtractor 8b. The register 8a is used to delay the picture image data by
one picture element and to supply to the subtractor 8b picture image data
f(i-1) one picture element before. The subtractor 8b performs the
subtraction operation shown in equation (1) in accordance with the two
inputs for outputting picture image data representing the primary
difference .DELTA.f(i).
FIG. 10 shows another example of the differentiating circuit 9 performing
the difference computation of the secondary difference shown in equation
(2) in which the picture image data f(i) sequentially derived out from
continuous picture element groups designated by the filter are applied to
a register 9a and a subtractor 9b. The purpose of the register 9a is to
delay the picture image data by one picture element and to supply the
picture image data f(i-1) one picture element before to a register 9c and
a shift register 9d. In the same manner as the register 9a, the register
9c delays the picture image data by one picture element for supplying to
the subtractor 9c picture image data f(i-2) two picture elements before.
The shift register 9d doubles the input picture image data f(i-1) by
shifting the same by one bit and supplies the doubled picture image data
to the subtractor 9b which subtracts picture image data f(i) from picture
image data 2f(i-1) for supplying the difference to a subtractor 9e. The
subtractor 9e subtracts the difference value from the subtractor 9e from
the picture image data outputted from register 9c for producing picture
image data representing the secondary difference .DELTA..sup.2 f(i-1).
When the absolute values of the differentiated outputs of the
differentiating circuits 8 and 9 exceed a given threshold value (i.e.
threshold value of brightness difference T; second data); it is judged
that a contour candidate point at which the brightness changes rapidly
exists, and at step 101, the direction of the variation of the brightness
(brightnss gradients; third date); is judged in accordance with the
positive or negative sign of the differentiated output.
If the scanning position thus checked is not the characteristic point of
the filter, the scanning position is moved to the right (X direction) by
several picture elements at step 102. The amount of this shift is
determined by the check ranges .DELTA.L.sub.1 and .DELTA.L.sub.2 of the
filter.
At step 103, a judgment is made as to whether the scanning position is at
the rightmost end of the scanning range or not. If the result of judgment
is NO, then at step 101, a judgment is made as to whether the present
position is the characteristic point of the filter or not. When the result
of judgment is NO, the processing described above is repeated.
In this manner, when the scanning position reaches the rightmost end of the
scanning range without finding the characteristic point of the filter, at
step 104, a judgment is made as to whether the scanning position is the
lowermost end or not.
When the scanning position is at the lowermost end, i.e., the position
P.sub.ED shown in FIG. 6, at step 105, it is judged that there is no
object to be detected in the picture and the processing of the picture
image is terminated. On the other hand, where the scanning position is not
at the lowermost end of the scanning range, the scanning position is
returned to the leftmost position, at step 106, the scanning position is
moved toward lower (Y)direction by a distance .DELTA.Y, as shown in FIG.
7. After that, the scanning is started again from that position to check
whether there is a characteristic point or not.
As the characteristic point of the filter is detected by scanning the
picture with the filter, at step 107, it is judged that an object was
detected at that scanning position.
It should be understood that the filter is not limited to that shown in the
above-mentioned embodiment in which ranges A.sub.1, A.sub.2,--, A.sub.n
arrayed along a single scanning line. As shown in FIGS. 11-14, ranges
(shown by circles) may be arranged radially around the reference position
(shown by a hatched circle). In this case, it would be desirable that each
range is arranged such that it crosses the contour line of the object to
be detected perpendicularly.
With the filter shown in FIG. 11, for example, filter parameters registered
in the memory 6 are as given in Table 2 below.
TABLE 2
______________________________________
First Data Second
Offset amount
Offset amount
Data Third
of one end of the other
(Brightness
Data
point from point from difference
(Bright-
reference reference threshold
ness
Contents
position (P)
position (P)
value) gradient)
______________________________________
Range
A.sub.1 X.sub.1, Y.sub.1
X'.sub.1, Y'.sub.1
T.sub.1 g.sub.1
(Lower
left)
A.sub.2 X.sub.2, Y.sub.2
X'.sub.2, Y'.sub.2
T.sub.2 g.sub.2
(Upper
left)
A.sub.n X.sub.n, Y.sub.n
X'.sub.n, Y'.sub.n
T.sub.n g.sub.n
(Right)
______________________________________
In the table, when the end points of each of the respective ranges are
selected to satisfy X.sub.n /X'.sub.n '=Y.sub.n /Y'.sub.n, a line
connecting the end points in each range becomes perpendicular to the
contour line of the object.
Where an object to be detected 10 is identified by a filter that specifies
ranges A.sub.1, A.sub.2 and A.sub.3 shown in FIG. 15, since the
characteristic point of the filter appears when the filter passes a
substantially central portion of the object, it would be impossible to
specify the correct position of the object. In such a case, the filter is
moved in the vertical direction (an auxiliary scanning direction) for
investigating a region containing the characteristic point of the filter
so as to specify the accurate position of the object. When an additional
judgment is made in which only when the traced length of the
characteristic point of the filter in the auxiliary scanning direction
exceeds a reference length, it is judged that the object 10 presents, so
that the object can be identified at a higher accuracy.
A method of identifying an object, a portion of its contour line being
parallel lines will now be described with reference to flow charts shown
in FIGS. 16, 17 and 18.
It is assumed that the object to be detected is a round rod 20 shown in
FIG. 19. In this case, a filter similar to that shown in FIG. 5 is used
wherein the center-to-center distance of the specific ranges A.sub.1 and
A.sub.2 corresponds to the diameter L of the round rod 20.
In the flow chart shown in FIG. 16, at steps 200-206, the cross-section of
the round rod 20 is searched. The processings executed at these steps are
identical to those executed at steps 100-106 shown in FIG. 8, so that the
description thereof will not be made.
A case wherein the cross section of the round rod 20, that is the
characteristic point of the filter, has been detected will be described.
In this case, the program is advanced to the flow chart shown in FIG. 17.
At first, as shown in FIG. 20, the filter position (scanning position) Pi
at which the filter characteristic point has been found is moved to a
center position Ps between two contour candidate points Pa and Pb, and at
step 211, this point Ps is stored as the tracing starting position Ps in
the auxiliary scanning direction. More particularly, when the filter
characteristic point thus detected is caused by desired parallel contour
lines, it may be considered that the filter characteristic points present
continuously in the vertical direction. For the purpose of checking such
continuity, the tracing starting position Ps is stored at first.
Then, the scanning position is moved upwardly by a distance S from the
starting position Ps, and at step 212, a check is made as to whether this
point is the filter characteristic point or not. Where the scanning
position is the filter characteristic point, that scanning position is
moved to the center position between two profile candidate points and the
scanning position is stored at step 213. Then, a check is made again as to
whether a point just above and spaced by a distance S from the cente point
is the filter characteristic point or not. In this manner, the tracing of
the filter characteristic point is continued until no more filter
characteristic point presents (see FIG. 20).
On the other hand, when a point just above and spaced from the scanning
position by a distance S is not the filter characteristic point, in other
words, the upward tracing of the filter characteristic point completes, at
step 214, the scanning position is returned to the tracing starting
position Ps. After that, at steps 215 and 216, the tracing of the filter
characteristic point is effected downwardly in the same manner as above
described until no more filter characteristic point presents.
As above described, when the tracing in the vertical direction of the
filter characteristic point starting from the tracing starting position
completes, the length of the tracing (the length of the parallel lines) is
determined from the uppermost point PV and the lowermost point PB (FIG.
19) among a number of scanning positions sequentially stored during the
tracing. Then at step 217, a judgment is made as to whether the traced
length is longer than a reference length or not. When the traced length is
shorter than the reference length, it is judged that the parallel lines
are not the desired ones. Then, the scanning for searching the filter
characteristic point is executed again according to the flow chart shown
in FIG. 16. When the traced length is longer than the reference length, at
step 218, a judgment is made whether the traced locus are arranged on a
substantially straight line or not.
As the method of judgment may be considered a linear approximation method
and a geometrical acceleration method. According to the former method, an
approximate straight line is obtained from all scanning positions stored
during the tracing and then a judgment is made as to whether substantially
all points are within a predetermined distance from the approximate
straight line. According to the latter method, twice differential
operations are sequentially made in the vertical direction with respect to
all scanning positions, and a judgment is made as to whether the twice
differentiated values are substantially zero or not. As a result of this
judgment, it becomes to identify two lines shown in FIGS. 21a and 21b and
the parallel lines shown in FIG. 21c. For the sake of description, two
lines shown in FIGS. 21a and 21b are herein called parallel lines.
When it is judged that the traced locus are not arranged in a straight
line, it is judged that they are not the desired parallel lines and the
scanning for searching the filter characteristic point is executed
according to the flow chart shown in FIG. 16. When it is judged that the
traced locus are on a substantially straight line, at step 230 shown in
FIG. 16, it is judged that the object has detected thereby ending the
processing of the picture image.
It is also possible to determine the inclination of the approximate
straight line. When the inclination is larger than a predetermined
reference value, the lines are judged that they are not parallel. With
this measure, it becomes possible to identify two types of parallel lines
shown in FIGS. 22a and 22b.
Instead of the tracing processing shown in FIG. 17, the tracing porcessing
according to the flow chart shown in FIG. 18 may be executed. Like the
flow chart shown in FIG. 7, the flow chart shown in FIG. 18 contains
processings of setting and changing the distance of movement S at the time
of tracing the filter position in the vertical direction. More
particularly, at step 220, the distance S of movement is set followed by
the tracing processings similar to those shown in FIG. 17. After judging
that the traced locus are arranged on a substantially straight line, at
step 221 shown in FIG. 18, a judgment is made as to whether the distance S
is sufficiently small or not. When the distance is sufficiently small, at
step 230 shown in FIG. 16, it is judged that the object has been detected,
thus ending the picture image processing. On the other hand, when the
distance S is not sufficiently small, at step 222, the distance S is
reduced and the filter characteristic point is again checked by using the
reduced distance S. The distance S set, the reference for judging whether
the distance S is sufficiently small, and the ratio of reduction the
distance S are determined by the fineness of the checking and the fineness
is determined by the fact that the picture im | | |
