JPH01298890A - Picture error evaluation method - Google Patents

Abstract

PURPOSE:To improve correspondence between a measure value of an error and a subjective deterioration quantity even if deterioration is caused locally by correcting the noise in response to the deterioration appearance area. CONSTITUTION:A difference detection circuit 3 is operated to take a difference between an original picture and a deteriorated picture and when the result is recorded in a frame memory 7, an error (e) of each picture element outputted from the difference detection circuit 3 is inputted simultaneously to an error calculation circuit 4, where noise N is obtained. On the other hand, a deterioration area extraction circuit 8 calculates an area where the error (e) appears and the picture element is separated into the error appearance picture element and the non appearance picture element by a threshold value processing in case of obtaining the error area to check the connection state between the error appearance picture elements. Then the deteriorated appearance area D collecting the error appearance picture elements close to each other is formed and an area rate A(D/(D+DN)) of the deterioration appearance area D is decided from the deterioration nonappearance area ND. Moreover, the noise NW corrected by a correction circuit 9 is displayed on a meter 5. Thus, the degree of the deterioration of the picture including the local deterioration is measured in a form well corresponding to the subjective deterioration of human being.

Description

Detailed Description of the Invention (Technical field to which the invention pertains) The present invention provides a method for measuring the degree of deterioration of an image such as a television image by measuring the difference (noise The present invention relates to an image error evaluation method for measuring image error (amount).

(Prior Art) Conventionally, the most common method for measuring the amount of image noise is to measure two images: an original image of sufficient quality in frame memory 1 and a deteriorated image in frame memory 2, as shown in FIG. This method calculates the root mean error (ruse).

That is, the luminance factor (x
t j) and the degraded pixel of frame memory 2 (, +, j'
) is added to the difference detection circuit 3, and the difference output, that is, the error e(x, J) of each pixel = I(x,
j) I''(x+j) is erroneous, and the noise amount N is displayed on the meter 5.

This method is easy to measure, and if the type of image quality deterioration is limited to random noise, etc., it is recognized as a measurement that corresponds well to subjective quality.

Furthermore, in order to improve the correspondence between the subjective degree of deterioration and the physically measured error even when the types of noise are different,
As shown in FIG. 4, a method is also used in which a filter 6 is used to obtain a corrected error based on the frequency brightness distribution.

By the way, the above-mentioned method is based on the premise that the entire screen is degraded to the same extent during boat fishing, as shown in FIG. Therefore, if the deterioration is localized as shown in Figure 6, even if the same type of deterioration has occurred and the error values measured using the measurement method described above are the same. However, there is a problem in that the degree of image deterioration subjectively viewed by a single person varies.

(Objective of the Invention) Therefore, in the present invention, in order to deal with the above-mentioned problems, the amount of noise is corrected according to the area where deterioration appears, and even when deterioration occurs locally, the measured value of error is The objective is to improve the correspondence between the amount of deterioration and the subjective amount of deterioration.

(Structure of the Invention) (Characteristics of the Invention and Differences from the Prior Art) In order to achieve the above object, the present invention detects the degree of deterioration based on the difference (amount of noise) between an original image of sufficient quality and an image in which deterioration has occurred. When measuring, the first step is to detect the amount of noise, the second step is to detect the area where the noise is occurring on the screen and find the area, and to make the area correspond to the found area. and a third step of correcting the noise amount measured in the first step, the most important feature is that a corrected noise amount that approximates human sensation is measured.

In conventional technology, the amount of noise is measured on the assumption that the entire screen is degraded to the same degree.However, the present invention can measure the amount of noise based on the area where the degradation appears even when the degradation is localized. The difference is that the amount of noise is corrected accordingly to improve the correspondence between the measured error value and the subjective amount of deterioration.

In order to investigate the dependence of subjective deterioration on area, the present invention
Deterioration area ratio A and the entire screen when the physical noise amount N is adjusted so that the subjective amount of deterioration is equal for the degradation of different deterioration area ratio A (area D of deteriorated part/total area) FIG. 2 shows an example of the measured relationship with the physical noise amount N (here rn+es) averaged by . In FIG. 2, both the vertical and horizontal axes are expressed in logarithms. From Figure 2, the relationship between the deterioration area rate A and the noise amount N is 1ogN = p ・logA + - 1ogk,
(1) However, it can be seen that P is a constant, that is, there is a relationship between N= and -A' (2). Here, in the example of Figure 2, p
is approximately 0.3 (>O), so the smaller the deteriorated area, the
This means that even if the error is small, deterioration can be felt. Therefore, when measuring the amount of noise, it is necessary to take such characteristics into consideration. Therefore, in order to obtain the amount of noise corrected to suit human senses, the following procedure may be used. First, the deterioration area rate A is determined, and then the noise amount N is corrected as follows: N, = N - A (3) If the corrected noise N, is determined, the subjective amount of deterioration and the physical measurement amount are calculated. corresponds well.

(Embodiment) FIG. 1 shows the configuration of an embodiment of an apparatus for carrying out the method of the present invention. In FIG. 1, frame memory 1 and frame memory 2 each record an original image and a degraded image. Here, in order to detect the difference between the original image and the degraded image, the difference detection circuit 3 is activated.

The results are recorded in the frame memory 7. At this time, the error e of each pixel output from the difference detection circuit 3 is simultaneously input to the error calculation circuit 4 to obtain the noise amount N. On the other hand, the degraded area extraction circuit 8 calculates the area where the error e appears. When calculating the error area, threshold processing is used to separate pixels that appear as errors and pixels that do not appear, examine the connection state between pixels that appear as errors, and create a degraded area that groups together adjacent pixels that appear as errors. The area ratio A (that is, D/(D + N D)) of the deterioration appearance area is determined from the deterioration non-appearance area ND.

Further, the correction circuit 9 performs the process shown in equation (3), and the corrected noise N is displayed on the meter 5.

In the above embodiment, the frame memories 1 and 2 are used to record the original image and the degraded image, but the frame memories may not be used and the data may be input to the difference detection circuit 3 in real time. Furthermore, although the original image has been given as an example with a pattern, it is also possible to use one without a pattern, such as a gray screen, and in that case, it is also possible to omit inputting the original image.

Furthermore, in the above embodiment, the root mean square error is used as the noise amount N, but other mean errors such as the absolute value mean error may be obtained. After performing the correction for 5 areas, the correction for 5 areas may be performed.

Furthermore, in the embodiment described above, the correction regarding the area was calculated using a power relational expression as shown in equation (3), but a more complicated correction may be performed using a table.

(Effects of the Invention) As explained above, the image error evaluation method of the present invention has the advantage that the degree of image deterioration, including local deterioration, can be measured in a manner that closely corresponds to the subjective amount of deterioration in humans. There is.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram of an embodiment of an apparatus for carrying out the method of the present invention, and FIG. 2 is an example of measuring the relationship between noise amount N and deterioration area ratio A when subjectively equal deterioration is given. FIGS. 3 and 4 are side views of each configuration of a conventional noise amount N measuring device, FIG. 5 is an example of an image with deterioration on both sides as a whole, and FIG. 6 is an example of an image with local deterioration. 1.2.7 Frame memory, 3 Difference detection circuit, 4 Error calculation circuit, 5 Meter,
6... Filter, 8... Deterioration area extraction circuit,
9...Correction circuit, D...Deterioration appearance area, ND・
...Area where no deterioration appears. Patent applicant: Nippon Telegraph and Telephone Corporation

Claims (1)

[Claims] When measuring the degree of deterioration based on the difference (amount of noise) between an original image of sufficient quality and an image in which deterioration has occurred, the first step is to detect the amount of noise, and the first step is to detect the amount of noise generated on the screen. A second process of detecting the area where the area is located and calculating its area, and a third process of correcting the amount of noise measured in the first process in accordance with the calculated area. An image error evaluation method characterized by measuring an approximate corrected noise amount.