JPS61201372A - Image processor - Google Patents

Abstract

PURPOSE:To restore and emphasize an image and to obtain useful information securely by performing spatial filtering with a different coefficient according to the place of an original image. CONSTITUTION:Different filter coefficients are stored in the 1st - the (N)th two-dimensional image memories 3A-3N, which are accessed with an address ad that a reference signal generating circuit 1 sends out together with a two-dimensional image memory 2 stored with the original image. Further, the reference signal generating circuit 1 sends out a selection signal (s) together with the address (ad) and a specific filter coefficient among the filter coefficients read out of the 1st - the (N)th image memories 3A-3N is selected corresponding to the address ad to multiply the original image by a multiplier 6; and an adder 7 sums up products, pixel by pixel, to perform the spatial filtering. Thus, the spatial filtering is performed differently, pixel by pixel, so the filtering processing is carried out according to the place, features of the original image.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to an image processing device that restores an incomplete original image and emphasizes characteristic parts.

[Technical Background of the Invention] Considering the process of obtaining the first image in image processing technology, there are various mechanisms that perform the process from taking the image to displaying it. Filtering is performed for repair purposes. Furthermore, filtering may be performed at individual intervals to compensate for the inability of human vision to sufficiently capture the information that an image is intended to convey.

By the way, when considering the constituent elements of a single image, the information contained in the image is enormous, and for example, different subjects often coexist in the periphery and center of the image.

Furthermore, an imaging device may have different frequency characteristics between its peripheral portion and its central portion, and such an example can be seen particularly in medical TV equipment and the like.

Conventionally, as an effective means in image processing technology, spatial filters have been applied to the original image to improve the frequency characteristics of the image itself and improve the information obtained from the original image. Depending on the nature of the original image, spatially stationary filter processing alone may not be enough to restore the image, so spatially unsteady filter processing is performed on the original image at high speed according to its characteristics to ensure sufficient image restoration and features. It was hoped that a partial extraction would be carried out.

[Problems with Background Art] Here, the conventional stationary filter (Space rn
An image processing technique using a variant filter will be described with reference to FIG.

The figure shows a memory 11 that stores images for each pixel.
In contrast, for example, an example is shown in which pixel A and pixel B are applied with a low-pass filter having exactly the same filter coefficients.

However, with such filter processing, only the same filter processing can be performed for each pixel, and in the example shown in Figure 4, since pixel A and pixel B are spatially separated, image restoration and enhancement should be performed differently. In reality, this may occur, but in this case, only isomorphic processing can be performed.

[Object of the Invention] The present invention has been made in view of the above circumstances, and uses a spatially unsteady filter (hereinafter referred to as [unsteady filter J (Space Variant Filter)] for the original image.
The object of the present invention is to provide an image processing device that performs image processing using (also referred to as )) and can realize restoration and emphasis of characteristic parts according to the characteristics of the original image.

[Summary of the Invention] A summary of the present invention for achieving the above object includes: a storage means for storing an original image to be filtered; an unsteady filter section for storing a plurality of unsteady filter coefficients; a reference signal generation circuit that provides a predetermined address system to the unsteady filter section; a calculation means that sequentially multiplies each pixel of the original image by a different unsteady filter coefficient; and an output of the calculation means for each frame of the original image. The image forming apparatus is characterized in that it has an accumulator that performs accumulation, and performs filter processing using unsteady filter coefficients on the original image.

[Embodiments of the Invention] The present invention will be described in detail below. First, to explain the unsteady filter which is the basic principle of the present invention, one
Address i of the original image stored in the dimensional image memory,
j (i is the horizontal address.

j is a vertical address; for example, if the image is 512x512 pixels, i=Q~511. j-0~511
It is. ) and calculate one pixel as D (t.

j).

If it is a stationary fimelta, a constant filter Go (
k, j! ), the pixel -D(i,j) after filtering the original image can be expressed by the following equation (1).

...(1) Here, if −≦≦ and −L<J≦L, then +2
1.21+1 gives the size of the filter Co (k, J). Also, i-k, j-J represent the translation address.

Next, if we express the unsteady filter in the same way, the pixel D (i, j) after filter processing in this case is as follows (21...
(a 2pi Here, in d, i (i, j) - Ci, j (k, j!
) XD (i-to, j-j!) ...(3
)It can be expressed as.

Here, k and l are coefficients, and a two-dimensional image memory group that constitutes a two-dimensional unsteady filter for all combinations of k and 1 is considered.

For example, when k=- and f=-L, Ci, H-K.

-L) filter coefficients enter the first two-dimensional image memory, and +1. When J = -L, Ci, j (+1 to -
．． -L) are entered into the second two-dimensional image memory.

Then, if −≦≦ and −L≦1≦L, the required number of two-dimensional image memories N is N= (2 + 1) x (
2L+1).

In this way, in each of the N two-dimensional image memories,
If the original image is subjected to filter processing using a non-stationary filter that stores non-stationary filter coefficients for all combinations of 1, it is possible to reliably achieve restoration and emphasis of characteristic parts according to the characteristics of the original image.

Next, an embodiment of the apparatus for realizing the above principle will be described with reference to FIG.

The device shown in the figure uses one system of address signals ad.

A reference signal generation circuit 1 that generates an address bias value and a selection signal S, and a storage means that is controlled by the address signal ad and selection signal S and stores an original image to be subjected to filter processing by inputting the address bias value. A two-dimensional image memory 2 (assuming addresses are represented by i and j) and coefficient matrices C1゜j(-, -L), Ci, j (+1.-L to -),...
....

1st to Nth two-dimensional image memories 3A, 3 in which Ci, j (K, L) are written and controlled by the address signal;
B, . . . , 3N and the first
~Nth two-dimensional image memory 3A, 3B.

..., an unsteady filter section 8 consisting of a selector 5 that selectively reads each coefficient matrix from 3N and sends it out, and the pixels of the original image from the two-dimensional image memory 2 and the coefficient matrix from the selector 8. A multiplier 6, which is an arithmetic means for performing multiplication of The accumulator 9 includes an adder 6 that sends the addition result to the calculation result storage memory 4 again.

Next, the operation of the apparatus of the above embodiment will be explained. The two-dimensional image memory 2 is controlled by an address signal ad and a selection signal S from a reference signal generation circuit 7, and converts addresses i and j to which an address bias value is added (to i-,
j-j! ) at pixel D(i-, j-, e
) is sent as one input signal of the multiplier 6.

At the same time, the selector 5 receives the selection signal S, selects the value of the coefficient matrix of any one of the outputs of the first to Nth two-dimensional image memories 3A to 3N controlled by the address signal ad, and multiplies it. It is transferred to the multiplier 6 as the other input signal of the multiplier 6.

As a result, the multiplier 6 divides the pixel o (r−, j−i >
The operation of multiplying by the coefficient matrix Ci,j (k, j!) is executed. That is, the calculation according to the above equation (3) is executed.

In this way, the multiplier 6 sequentially applies coefficients to each pixel D(i, j) to the coefficient matrix Ci of all combinations of 1.
, j(k, fl), and sends these multiplication results to the adder 7 of the accumulator 9.

The adder 7 executes the calculation based on the above equation (4) for each frame of the original image. That is, the coefficient j! Every time the value of changes once, the multiplication result by the multiplier 6 and the output of the calculation result storage memory 4 that stores the addition result by the adder 7 are added pixel by pixel, and this result is returned to the calculation result storage memory. Store in 4.

An adder 7 adds all combinations of 1 to the coefficients. The result calculation result storage memory 4 is stored in the above (
4) D(i, j) of the equation is stored, and the calculation by this device is completed.

In this way, as shown in Fig. 2, for pixel A of the original image, the coefficient matrix C^ as shown in the same figure is
Pixel B will be multiplied by the coefficient matrix CB, and pixel A and pixel B can be subjected to repair 1 enhancement.

FIG. 3 shows an apparatus according to another embodiment of the present invention. Components having the same functions as those of the apparatus shown in FIG. 1 are given the same reference numerals, and detailed explanation thereof will be omitted.

The difference between the apparatus shown in FIG. 1 and the apparatus shown in FIG. 1 is that the unsteady filter section 8 is constituted by a coefficient signal generator 8A.

This coefficient signal generator 8A generates j! are parameters and addresses i and j are variables, and in the filter coefficient Ci,j (k, 12), for one set, Ci,j is Ci,
It is given by a single algebraic number related to i and j, such as j-f (j, j).

In this case, a set of 1 for a certain coefficient corresponds to each of the coefficient matrices Ci, j (k, j!), so the first to Nth two-dimensional image memories 3A to
It can perform the same function as 3N and selector 5.

The present invention is not limited to the embodiments described above, and various modifications can be made within the scope of the invention.

For example, in the device shown in FIG.
Addresses i and j in the two-dimensional image memory 2 are converted into conversion addresses i- and j-1 by the address bias value and the address bias value. , j and an address signal that gives j-z to the conversion address i- are generated separately, and the former is sent to each two-dimensional image memory and calculation result storage memory of the unsteady filter section, and the latter is sent to the secondary The same implementation is possible by giving each to the image memory.

[Effect of the invention] According to the present invention detailed above, the location of the original image.

It is possible to provide an image processing device that can realize unsteady filter processing according to its characteristics, perform restoration and emphasis on an image according to its characteristics, and reliably obtain useful information.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment device of the present invention;
FIG. 2 is an explanatory diagram showing an outline of unsteady filter processing in the same device, FIG. 3 is a block diagram showing the configuration of another embodiment of the device of the present invention, and FIG. 4 shows an outline of conventional steady filter processing. It is an explanatory diagram. DESCRIPTION OF SYMBOLS 1... Reference signal generation circuit, 2... Two-dimensional image memory, 6... Multiplier, 8... Unsteady filter section, 9...
·accumulator.

Claims (4)

[Claims] (1) A storage means for storing an original image to be filtered, an unsteady filter section for storing a plurality of unsteady filter coefficients, and a reference signal generation circuit for providing a predetermined address system to the storage means and the unsteady filter section. , an arithmetic means for sequentially multiplying each pixel of the original image by a different non-stationary filter coefficient, and an accumulator for accumulating the output of the arithmetic means for each frame of the original image. An image processing device characterized by performing filter processing using filter coefficients. (2) The image processing apparatus according to claim 1, wherein the unsteady filter section is a coefficient signal generator that generates unsteady coefficients. (3) The reference signal generation circuit generates one system of address signals and an address bias value, and the storage means creates a new address based on the address signal and address bias value, and generates a new address of the original image. 2. The image processing device according to claim 1, wherein the image processing device transmits the pixel values of . (4) The image processing device according to claim 1, wherein the reference signal generation circuit is configured to provide two relatively different address systems to the storage means and the unsteady filter section.