Image signal filtering - Patent Review 4817180

Claims

We claim:

1. A method of producing an unsharp filtering signal for an image signal representing image data obtained from a scanning process in main and sub scanning directions comprising the steps of:

(a) filtering said image data in at least one direction corresponding to one of said main and sub scanning directions,

(b) said filtering step comprising processing image signals associated with a matrix of pixels,

(c) performing a cumulus computation operation on a pair of processed image signals P.sub.n and Q.sub.n obtained from said processing step where n is an input order or computation number of the signals, said cumulus computation defined by

R.sub.n =R.sub.n-l P.sub.n -Q.sub.n

where R.sub.n-1 is a result of an immediately preceding cumulus computation, and

(d) utilizing a signal resulting from the cumulus operation performed on the filtered signals in an image generation process.

2. A method as recited in claim 1 wherein said processing step comprises the steps of selecting a predetermined number K of image signals d.sub.i (i=K, . . . , l) out of 2K image signals obtained from the scanning process representing a scan in said one of said main and subscanning directions,

summing the selected number of signals to produce said first signal, and

summing the remaining K image signals to produce said second signal.

3. A method as recited in claim 2 wherein said filtering step comprises the further step of processing the image data in the other of said main and sub scanning directions by obtaining from said signals filtered in said one direction further first and second signals and subjecting said further first and second signals to a further cumulus computation.

4. A method as recited in claim 2 wherein said 2K image signals represent two data sequences, K computations apart, and said first selecting step comprises the step of selecting substantially all of K sequential signals in a scan in one direction, and said remaining K image signals represent a sequence forming the first signal of a K.sup.th previous computation,

said second summing step providing a sum of a sequence forming a first signal of a K.sup.th previous computation.

5. Apparatus for filtering an image signal in a main scanning direction comprising storage means for receiving a scanning signal in the main scanning direction

means for performing a pair of cumulus computations, defined as providing an n.sup.th result R.sub.n =R.sub.n-1 +P.sub.n -Q.sub.n for an n.sup.th computation where R.sub.n-1 is a result of an immediately preceeding computation and where P.sub.n and Q.sub.n are image signals input to said means for performing said cumulus computations, and

means for performing a third cumulus computation on results of said pair of said cumulus computations, to provide a filtered signal output representing a filtering of said scanning signal in the main scanning direction,

said means for performing a pair of cumulus computations including a structure for performing a first cumulus computation connected for receiving positive and negative inputs from an element of said scanning signal and from an element of said scanning signal delayed by K scan elements, and including a structure for performing a second cumulus computation connected for receiving positive and negative inputs from said element of said scanning signal delayed by K scan elements and from an element of said scanning signal delayed by 2K scan elements,

outputs representing said first and second cumulus computations providing positive and negative inputs to said means for performing said third cumulus computation.

6. Apparatus as recited in claim 5 wherein said storage means comprises shift register means having 2K+1 storage stages.

7. Apparatus as recited in claim 5 wherein said means for performing a third cumulus computation includes adding means receiving positive and negative inputs, said adding means having a further positive input connected for receiving after a single delay period an output thereof.

8. Apparatus as recited in claim 7 wherein said means for performing a pair of cumulative computative computations includes a pair of structures substantially identical to said means for performing said third cumulative computation.

9. Apparatus as recited in claim 7 wherein said means for performing a pair of cumulative computations includes first cumulative computing means comprising adding means having a pair of positive inputs and a negative input, said adding means connected for receiving at one of said positive inputs an element of said scanning signal, and connected further for receiving at said negative input an element of said scanning signal after delay by K scan elements, and further connected for receiving at the second positive input an output thereof delayed by one delay period, and

said second cumulative computing means comprises shift register means having K stages, a first stage connected to receive an output from said first cumulative computing means, and a K.sup.th stage connected to provide an output to a negative input of said adding means of said means for performing said third cumulative computation.

10. Apparatus as recited in claim 9 wherein said storage means comprises further shift register means having K+1 stages, said first stage providing said first positive input to said first cumulative computing means and said K+1.sup.th stage providing said negative input to said first cumulative computing means.

11. Apparatus as recited in claim 6 wherein said shift register means a plurality of line memory storage means provided in cascade arrangement, each of said line memory storage means incuding 2K+1 storage cells for elements of one scanning line.

12. Apparatus as recited in claim 5 wherein said means for performing a third cumulus computation includes adding means receiving positive and negative inputs, said adding means having a further positive input delayed by one line scan.

13. Apparatus as recited in claim 12 wherein said structures for performing said first and second cumulus computations are substantially identical to said means for performing said third cumulative computation.

14. Apparatus as recited in claim 13 wherein said plurality of line memory storage means comprises 2K+1 line memory storage means, and an output of a first line storage means and an output of a K+1.sup.th line storage means providing positive and negative inputs to said structure for performing said first cumulative computing means, an output from said K+1.sup.th line memory storage means and an output from said 2K+1.sup.th line storage means providing respectively positive and negative inputs to said structure for performing said second cumulus computation.

15. Apparatus as recited in claim 12 wherein the structure for performing the first cumulus computation is substantially identical to said means for performing the third cumulus computation and the structure for performing the second cumulus computation comprises K-1 line memory storage means connected in cascade with the line memory storage means at the output of said structure for performing the first cumulus computation, an output of the last of said K-1 line memory storage means providing a negative input for said means for performing a third cumulus computation.

16. Apparatus as recited in claim 8 wherein said storage means comprises shift register means having 2K+1 storage elements further comprising means for imposing a weight coefficient increasing from edge image signals to a K.sup.th image signal, including

fourth and fifth cumulus computation means each substantially identical to said means for performing a third cumulus computation,

multiplying means connected to a K+1th stage of said shift register means for multiplying by K a signal stored therein, said multiplying means and the output of said adding means in said structure for performing said first cumulus computation provided to said fourth cumulus computation means as negative and positive inputs, respectively,

the output of said multiplying means and of said adding means of said structure for performing said second cumulus computation provided as positive and negative inputs to said fifth cumulus computation means, first and second subtracting means connected for subtracting outputs of said structures for performing said first and second cumulus computations from multiples of outputs of said fourth and fifth cumulus computation means said first and second subtracting means providing positive and negative inputs to said means for performing said third cumulus computation, said output of said means for performing said third cumulus computation providing an unsharp signal for input to a third subtracting means for subtraction from a multiple of a sum of an output of said fourth cumulus computation means and a difference between the output of said fifth cumulus computation means and the output of said structure for performing said second cumulus computation, to provide an unsharp signal weighted by weight coefficients varying in accordance with a quadratic function increasing from edge image signals to the K.sup.th image signal.

17. An apparatus for filtering an image signal in reproducing images by means of an image reproduction system, in which a cumulus computation expressed by:

R.sub.n =R.sub.n-l +P.sub.n -Q.sub.n

(where n is an input order of each of the signals to the above computation) wherein P.sub.n and Q.sub.n are image signals obtained in synchronism with an input scan to be subjected to the above computation and R.sub.n-1 is carried out by an apparatus comprising:

(a) a memory means for successively storing and shifting a sequence of a number of image signals comprised of one of a main or a sub-scanning line obtained from an input scanning meas for processing in synchronism with the input scanning means;

(b) a first summing means for obtaining in synchronism with the input scanning means a first sum signal U.sub.la expressed by: ##EQU31## by summing a predetermined number (K) of image signals d.sub.i (i: K, K-1, . . . 1) of 2K image signals d.sub.i (i: K, . . . , 0, 1, -1, . . . , K+1) in a sequence stored and shifted in accordance with the input scanning means;

(c) a second summing means for obtaining in synchronism with the obtaining of the first sum signal by said first summing measn a second sum signal U.sub.lb expressed by: ##EQU32## by summing the remaining (K) image signals; and (d) a cumulus computation means for obtaining a filtered signal U.sub.2 expressed by: ##EQU33## according to the input scanning means by subjecting the first and second sum signals U.sub.la and U.sub.lb to said cumulus computation as respective addition and subtraction signals for the (2K-1) image signals d.sub.i (i: K, . . . -K-1) in a sequence.

18. Apparatus as recited in claim 11 wherein said means for performing a third cumulus computation includes adding means receiving positive and negative inputs, said adding means having a further positive input delayed by one line scan.

FIELD OF THE INVENTION

The present invention relates to a filtering process of an image signal, and more particularly, to a method of and apparatus for filtering an image signal in producing an unsharp signal used for a process of detail emphasis.

BACKGROUND OF THE INVENTION

A detail (sharpness) emphasis process of an electronic image reproduction is achieved by carrying out a computation S+k(S-U) (where k is a coefficient) wherein S is a sharp signal representing the density of a pixel presently being processed, while U is an unsharp signal representing the average density of a plurality of pixels surrounding the present pixel. In this regard, an analog unsharp signal U can be optically obtained by using a scanning beam the aperture of which is broader than that used for obtaining the sharp signal S. However this method has a fatal drawback of requiring an independent optical system for obtaining the unsharp signal U, other than that for the sharp signal S. In addition, the beam aperture for the sharp signal S is required to be varied according to the sort of an image (drawing or gradated image) from which an image signal is obtained, a magnification ratio to be employed, and a screen ruling in reproducing halftone dots. Therefore, inevitably the aperture for the unsharp signal U most be similarly varied. Such a system is complicated and the operation procedures are time-consumptive.

For the purpose of solving these defects, Japanese patent application No. 54-82571 discloses a method of a digital detail emphasis process, and U.S. patent application Ser. No. 573,967 discloses an improvement of the above identified method.

In either method, an averaging circuitry or a combination of a self-multiplier and an adder is employed to process image signals of a group of pixels arranged in accordance with the scanning order, to each of which is applied a weighting coefficient.

Since the image signal filtering circuitry (an unsharp signal producing circuitry) employs multipliers or adders corresponding to the aperture size (physical mask size) of the unsharp signal, the greater the aperture for obtaining the unsharp signal, the more multipliers or the adders that must be employed. Furthermore, either method is accompanied by an inconvenience that identical circuits are necessary for each change of a weight coefficient to be imposed on the image signal.

SUMMARY OF THE INVENTION

Accordingly, it is an object of the present invention to provide a method of and apparatus for filtering an image signal free of the above-mentioned drawbacks.

Another object of the present invention is to provide such a method and apparatus free of the need for increasing the number of elements, such as adders, comprised in computation devices to cope with the increase of the number of image signals (corresponding to aperture size in analog signal processing) to be subjected to the image signal filtering process at a time.

A further object of the present invention is to provide such a method and apparatus capable of freely varying the number of image signals to be subjected to the image signal filtering process at a time.

In order to implement the above objects, the present invention adopts the following techniques.

It is prepostulated that a cumulus computation of the present invention means that two signals P.sub.n and Q.sub.n and the previous computation result signal R.sub.n-1 (where n is the input order or computation number of the signals) are combined in a computation according to the equation:

R.sub.n =R.sub.n-1 +P.sub.n -Q.sub.n.

It is stipulated in this specification that the aforesaid computation of the cumulus computer is described as "the signals P.sub.n and Q.sub.n are put to a cumulus computation" hereinafter for the sake of convenience.

An unsharp signal is obtained by processing image signals of a matrix of pixels (arranged in two dimensions), when the image data is firstly filtered in one of the one-dimensional main and the sub-scanning direction factors, and then the one-dimensionally filtered signal is further processed in the other direction.

In regard to the one-dimensional filtering process, first a certain number (k) of image signals d.sub.i, (i=k, . . . 1) out of 2k image signals in a sequence successively obtained from an input scanning are summed to produce a first sum signal U.sub.1a according to an equation: ##EQU3## meanwhile, the other k image signals d.sub.i (i=.0., . . . -k+1) are summed to produce a second sum signal U.sub.1b according to an equation: ##EQU4## Secondly, the first and the second sum signals are put to a cumulus computation to obtain a filtered signal U.sub.2 expressed by an equation: ##EQU5##

The first and the second sum signals can be obtained in accordance with any of the following methods.

In one method, firstly (2k+1) image signals are temporarily stored in a signal storage means such as a shift register, and then the first image signal and the (k+1)th image signal are out of the (2k+1) image signals are put to a cumulus computation, as addition and subtraction signals respectively, to obtain the first sum signal. Meanwhile, the (k+1)th image signal and the (2k+1)th image signal are put to another cumulus computation, as addition and subtraction signals respectively, to obtain the second sum signal.

In another method, firstly 2k image signals are temporarily stored in a signal storage means such as a shift register, and then the first through kth image signals out of 2k image signals are input to adders in a tree arrangement of several stages to be successively added to to an adjoining image signal until the first sum signal is formed. Meanwhile, the (k+1)th through 2kth image signals are input to other adders of the same construction as above until the second sum signal is formed.

In view of the fact that the second sum signal is delayed in phase by k computations from the first sum signal, the first sum signal can be delayed in phase by k computations to cause the second sum signal to be output in synchronism with the present first sum signal. In this case, the first sum signal is obtained first by temporarily storing (k+1) image signals d.sub.i in a first signal storage means, and then by putting the first image signal d.sub.k and the (k+1)th signal d.sub.O to a cumulus computation as respective addition and subtraction signals. The first sum signal can also be obtained by successively carrying out additions of several stages between adjoining two of k image signals temporarily stored in the signal storage means. The thus-obtained first sum signal is input to a second signal storage means having a siganl shifting function and then output as the second sum signal after k computations.

The aforesaid equation (31) can be obtained by a combination of the equations: ##EQU6## More particularly, the first sum signal U.sub.1a is first obtained according to one of the said methods, and then the first sum signal U.sub.1a and signal kd.sub.O obtained by multiplying the (k+1)th image signal d.sub.O by k are put to a cumulus computation, respectively as addition and subtraction signals, to obtain the filtered signal U.sub.2a.

The filtered signal U.sub.2b can also be obtained by putting the first sum signal U.sub.1a and a signal kd.sub.k obtained by multiplying the first image signal d.sub.k by k to a cumulus computation, respectively as subtraction and addition signals, to obtain a signal U'.sub.2b expressed by an equation: ##EQU7## which is advanced in phase by k computations from the filtered signal U.sub.2b, and by delaying the signal U'.sub.2b in phase by k computations.

Since the equation (31) can be expressed by an equation:

U.sub.2 =U.sub.2a +U.sub.2b -kd.sub.1 - - - ( 34),

the filtered signal U.sub.2 corresponding to (2k-1) image signals represented by the equation (31) can be obtained by carrying out the equation (34) according to the filtered signals U.sub.2a and U.sub.2b.

Since the signal U.sub.2b -kd.sub.1 comprised in the equation (34) can also be obtained by shifting by k computations the resultant of a computation kU.sub.1a -U.sub.2a using the filtered signals U.sub.1a and U.sub.2a, the filtered signal U.sub.2 represented by the equation (31) can be obtained by summing the signal U.sub.2b -kd.sub.1 and the filtered signal U.sub.2a.

The filtered signal U.sub.2 represented by the equation (31) can be obtained for image signals comprised in either the main scanning or the sub-scanning direction. To obtain the signal U.sub.2 filtered in the main scanning direction factor, a shift register may be used comprising a number of elements corresponding to the number of the image signals to be subjected to the filtering process.

To obtain the signal U.sub.2 filtered in the sub-scanning direction factor, a necessary number of line memories (or shift registers) are arranged in cascade, each unit of which can store image signals of one scanning line, thereby the image signals comprised in the sub-scanning direction are simultaneously output from each unit of the line memories (or shift registers).

As mentioned before, there must be carried out signal filtering processes in both the main and the sub-scanning directions to obtain an unsharp signal. For that reason, an image signal d.sub.i is filtered in one of the main and the sub-scanning direction factors, and then the obtained filtered signal U.sub.21 is fu