JPH05176179A - Method and device for estimating picture data subject to compression/expansion processing - Google Patents

Abstract

PURPOSE:To provide a method and a device in which quality of a compressed/ expanded picture is automatically and quantitatively evaluated apart from human subjectivity. CONSTITUTION:Picture data of an original picture are inputted to a 1st input means (a) and picture data resulting from applying compression/expansion processing to the original picture are inputted from a 2nd input means (b) to a spatial frequency spectral arithmetic operation means (c). The spatial frequency spectral arithmetic operation means (c) obtains a spatial frequency spectrum from each of the inputted picture data. A spatial frequency spectral extract means (f) extracts only a spatial frequency spectrum in a specific spatial frequency area giving effect onto a human visual characteristic, a spatial frequency spectral comparison means (d) compares them, an evaluation result output means (e) evaluates the picture deterioration and outputs the result based on the obtained comparison result.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of evaluating an image subjected to compression / expansion image processing, and more particularly to a method of evaluating a so-called multi-valued image having gradation.

[0002]

2. Description of the Related Art In the field of image processing in recent years, since image data is processed using a computer, input analog data is often A / D converted and treated as digital data. However, since the image is handled in units of fine pixels, the amount of data becomes enormous. For example, if an image of A4 size is expressed by four color components at a resolution of about 15 lines / mm, a data amount of 56 MByte is required. .. Therefore, the image data is usually compressed when storing or transmitting the image data. Image compression reduces the amount of information by quantizing spatial frequency components of image data by allocating a large number of bits for frequency components with a high appearance rate and a small number of bits for frequency components with a low appearance rate. It is a thing.

However, the degree of image deterioration is
This is performed by relative comparison with the image before decompression processing, but the reproducibility deteriorates in the frequency part to which a smaller number of bits are assigned during data compression, so the same compression method (device)・ Even if compression is performed at the same compression ratio, if the spatial frequency distribution differs for each target image,
The degree of deterioration with respect to the original image will also differ. Also,
Such deterioration may occur in a region that appears and a region that does not appear in the same image. In this way, since the degree of quality deterioration changes for each target image, it is not possible to use a specific chart or the like for the evaluation of compressed / decompressed images, and these evaluations are generally performed by visual inspection. It was target.

[0004]

However, since the evaluation of the compressed / decompressed image as described above is an evaluation based on human subjectivity, there are individual differences, and there is a possibility that the evaluation standard may differ from person to person. In addition, since the evaluation result cannot be captured quantitatively, even if it is necessary to change (drop) the compression rate and recompress again, for example, if there is deterioration beyond the standard as a result of performing compression / expansion, It was not possible to obtain a guideline for how much the compression ratio should be changed so that the deterioration can be kept within the standard. Therefore, an object of the present invention is to provide a method and an apparatus for automatically and quantitatively evaluating the quality of compressed / decompressed images without depending on human subjectivity.

[0005]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and the invention according to claim 1 is a method for evaluating compressed / decompressed image data, which is before compression. A feature is that the spatial frequency spectrum of the image data of the original image is obtained, the spatial frequency spectrum of the image data of the image after the compression / expansion processing is obtained, and the image data is evaluated by comparing the two spatial spectra. This is a method for evaluating compressed / decompressed image data.

According to a second aspect of the present invention, in the method of evaluating image data that has been compressed / decompressed, the spatial frequency spectrum of the image data of the original image before compression is obtained,
The spatial frequency spectrum of the image data of the image after the expansion processing is obtained, the spatial frequency spectrum of the frequency region that affects human visual characteristics is extracted from the two spatial frequency spectra, and the extracted spatial frequency spectra are compared. This is a method for evaluating compressed / decompressed image data.

According to a third aspect of the present invention, there is provided a first input means a for inputting image data of an original image, a second input means b for inputting compressed / decompressed image data, and the second aspect. Spatial frequency spectrum calculating means c for obtaining a spatial frequency spectrum of image data input from one input means
And a spatial frequency spectrum comparing means d for comparing the two spatial frequency spectra, and an evaluation result output for evaluating image deterioration based on the comparison result obtained by the spatial frequency spectrum comparing means and outputting the evaluation result. Means e
An apparatus for evaluating compressed / decompressed image data, comprising:

According to a fourth aspect of the present invention, there is provided a first input means a for inputting image data of an original image, a second input means b for inputting compressed / expanded image data, and the second Spatial frequency spectrum calculating means c for obtaining a spatial frequency spectrum of image data input from one input means
A spatial frequency spectrum extracting means f for extracting a spatial frequency spectrum of a predetermined frequency region from the two spatial frequency spectra, and a spatial frequency spectrum comparing means d for comparing the obtained spatial frequency spectra of the two predetermined frequency regions. And a compression / expansion processed image, which comprises an evaluation result output means e which evaluates image deterioration based on the comparison result obtained by the spatial frequency spectrum comparison means and outputs the evaluation result. This is a data evaluation device.

[0009]

According to the compressed / decompressed image evaluation method of the present invention,
Image data of the original image (image before compression) and compressed to the original image
A spatial frequency spectrum is obtained from the decompressed image data after the decompressed image by two-dimensional Fourier transform or the like. Then, the image is evaluated by comparing the spatial frequency spectra of both. At this time, if the components of the spatial frequency spectrum in the frequency domain that affect the human visual characteristics are extracted and only the spatial frequency spectra of these components are compared, only the image deterioration that is a problem for human vision is found. It becomes possible.

Further, in the apparatus for evaluating compressed / decompressed image data of the present invention, the image data of the original image is input to the first input means a, while the original image is input from the second input means b. The image data after the compression / expansion process is performed is input. For the image data input in this way,
The spatial frequency spectrum calculation means c obtains the spatial frequency spectrum using a technique such as two-dimensional Fourier transform. From the obtained spatial frequency spectra of the two images, the spatial frequency spectrum extraction means f extracts only the spatial frequency spectrum of the specific spatial frequency region that affects the human visual characteristics. The spatial frequency spectrum of the specific frequency region thus extracted is compared by the spatial frequency spectrum comparing means d, and based on the obtained comparison result, the evaluation result outputting means e evaluates the image deterioration and outputs the result.

Although the "means for solving the problem" and the "action" have been described with reference to the symbols shown in FIG. 1, the invention is not limited to the configuration shown in FIG. Of course.

[0012]

An embodiment of the present invention will be described in detail with reference to the drawings. FIG. 2 is a block diagram showing a configuration example of the production plan creation device according to the present invention. An image input device 1 such as a scanner is for reading an original image such as a film. The type of original image is not limited to this,
An image generated by computer graphics or the like may be input, or an image from an optical disc or the like may be captured. Reference numeral 2 denotes a computer such as a workstation, which compresses and decompresses images.

Further, a function of obtaining a spatial frequency spectrum by performing a two-dimensional Fourier transform or the like on image data, a function of extracting only a spectrum of a specific spatial frequency region from the spatial frequency spectrum, and a comparison of two spatial frequency spectra It has a function to perform and a function to evaluate the image deterioration from the comparison result. Reference numeral 3 denotes an instruction device including a keyboard or the like for instructing various processes of the computer 2 and setting a frequency region of a spatial frequency spectrum. Reference numeral 4 denotes a display device including a CRT monitor and the like, which displays an input image, an image subjected to compression / expansion processing, an evaluation result, and the like. An image output device 5 including a color printer outputs the input image and the compressed / decompressed image as a hard copy. Reference numeral 6 is a storage device including a hard disk, an optical disk, and the like. The original image data and the image data subjected to the compression / expansion processing are input into the storage device 6. Further, the storage device 6 stores data for making a predetermined evaluation on the result of comparison of the spatial frequency spectra.

Next, the operation of the present invention will be described with reference to the flowchart shown in FIG. 4 and FIG. Image compression / decompression image First, the image data of the original image 11 is fetched from the image input means 1 (step SP1). An image captured from a photograph, a film or the like is A / D converted by the computer 2 and stored in the storage device 6 as image data 12. When it is desired to confirm the contents, it can be output as a monitor image or a hard copy from the display device 4 or the image output means 5.

Next, the computer 2 performs compression / expansion processing on the input image data (step SP).
2). Although many methods of image compression algorithms have been proposed, a method using DCT (discrete cosine transform) is generally used. As an example, the JPEG method, which is a standard for image compression using DCT, will be described with reference to FIG. First, the image is divided into blocks of 8 × 8 pixels, DCT is applied to each of the divided blocks, orthogonal encoding is performed, and frequency characteristics are obtained. Next, the coded image is quantized with a predetermined quantization table. Data compression is performed by performing entropy coding on the quantized data using Huffman coding. In order to decompress the compressed data, the reverse process may be applied to the compressed data. The image data 13 that has been compressed / decompressed as described above is stored in the storage device 6. Further, the display device 4 or the image output means 5 can display or output a hard copy.

Calculation of spatial frequency spectrum A spatial frequency spectrum is obtained for the original image and the two image data that have been subjected to compression / expansion processing (step S).
P3). There are several methods for obtaining the spatial frequency distribution of a two-dimensionally distributed signal, but the Fourier transform is often used because of the ease of processing. Two-dimensional Fourier transform F for discretely sampled image data L (x, y)
(U, v) is represented by the following equation.

[0017]

[Equation 1] Therefore, u and v spectra (power components) are expressed by the following equation.

[0018]

[Equation 2] As described above, the spatial frequency spectra 14 and 15 of the original image and the two images compressed and expanded are obtained.

Comparison of Spatial Frequency Spectra Next, the two spatial frequency spectra 14 and 1 obtained above are calculated.
Spatial frequency 6 cycles / degree from 5
The spatial frequency spectrum of the neighboring region is extracted (step SP4). For example, 5-10 cycles / degree
The spatial frequency spectrum in the frequency domain in the range is extracted. Since image compression / decompression is realized by reducing the amount of information in a visually insignificant portion (usually a high frequency component), it can be sufficiently predicted that the spatial frequency spectrum will change. Therefore, by directly comparing the spatial frequency spectra obtained as described above, it is possible to examine the deterioration of the image in all the spatial frequency regions of the image, but also for the deterioration of the portion that does not visually affect humans. It may be extracted. In this embodiment, attention is paid to a visually significant portion in the change of the spatial frequency spectrum.

Typical image quality deterioration factors in image compression / expansion are block distortion in which a discontinuous portion appears in a pattern portion (low frequency region), which should be originally smooth, and ghost that should not be present in the edge portion of the pattern. There are false contours that appear.
The block distortion is a change in spectrum in the low frequency region. Further, the false contour appears linearly in the vicinity of a portion where the amount of change in the luminance distribution in the image is large, and changes in the high frequency component in the spatial frequency domain. On the other hand, human visual characteristics are known to have a response peak around a spatial frequency of 6 cycles / degree ("TWO-DIM").
ENSIONAL SIGNAL AND IMAGE
PROCESSING "<JAES. LIM 1989
Published by PRENICE HALL> P433,
"FUNDAMENTALS OF DIGITAL
IMAGE PROCESSING "<ANIL K.
By JAIN 1989 PENTICE HALL
If the image deterioration occurs in this spatial frequency region, it becomes particularly noticeable to humans.

FIG. 6 shows an example of a spatial frequency spectrum of an original image and an image obtained by subjecting the original image to compression / expansion processing. Comparing the spatial frequency spectrum 21 of the original image and the spatial frequency spectrum 22 of the image subjected to the compression / expansion processing, the spatial frequency region 23 shows the influence of block distortion, and the spatial frequency spectrum 24 shows the influence of false contour. However, the image deterioration in this example is not limited to the spatial frequency region 25 (spatial frequency 6 cycles / degree), which has a strong influence on human vision.
Since it appears in the vicinity of), it becomes a particular problem when seen by humans. On the other hand, if these deteriorations occur in areas other than the spatial frequency region 25, there is little effect on humans (it is difficult for humans to judge the deteriorations), so it can be said that there is no problem in quality.

The present invention pays attention to this point, and
The spatial frequency spectrum of the specific region 25 is extracted from each of the images subjected to the decompression processing, and the comparison is performed by emphasizing the spatial frequency spectrum of the two images (step SP).
5).

The method of comparing two spatial frequency spectra is, for example, that the spatial frequency spectrum is defined by a predetermined frequency region 25.
Alternatively, the area formed by the spatial frequency spectrum may be obtained by integrating over the entire frequency region, and the difference between the two may be calculated. Further, as another method, the correlation coefficient of the two spatial frequency spectra may be obtained.

Image Evaluation Obtained as described above,
The image is evaluated based on the comparison result (step SP
6). Specifically, an evaluation coefficient (evaluation point) corresponding to a value (or a value of a correlation coefficient) obtained by subtracting the integral value of each spectrum of the original image and the image subjected to the compression / expansion processing is determined in advance and stored in the storage device 6. The evaluation coefficient corresponding to the difference value (or the correlation coefficient) is called and displayed on the display device 4 (step SP7). Therefore, the operator can quantitatively determine whether or not the deterioration of the image due to the image compression / expansion this time is within a certain standard based on the value of the evaluation coefficient.

In the present embodiment, only the spectrum of the spatial frequency domain 25 is used as the spatial frequency spectrum used for comparison, but the spatial frequency spectrum of all the frequency domains of the image may be compared, or another method may be used. As a method, a method of performing a predetermined weighting correction on the extracted frequency region 25 and then comparing the two spatial frequency spectra in the entire frequency region may be considered.

In the present embodiment, the original image and compressed
Although comparison of decompressed images is performed in the same computer 2, the invention is not limited to this. If the image is to be transferred via a communication line, the transmitting side sends the spatial frequency spectrum data of the basic image together with the image data, and the receiving side compares the two by determining the spatial frequency spectrum of the expanded image, If the deterioration is above the standard, request transmission again, or feed back the spatial frequency spectrum data of the decompressed image on the receiving side to the sender to evaluate, and if the deterioration is above the standard, retransmit. It is also possible to do. Further, the compression rate can be determined according to the value of the evaluation coefficient when the compression is performed again.

[0027]

As described above, according to the present invention, the evaluation of the image quality of the compressed / decompressed image is evaluated by the human by comparing the spatial frequency spectra of the original image and the image obtained by compressing / decompressing the original image. It can be done automatically regardless of the result, and the result can be expressed quantitatively.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a compression / decompression-processed image data evaluation apparatus according to the present invention.

FIG. 2 is a block diagram showing a schematic configuration of a device for evaluating compressed / decompressed image data according to an embodiment of the present invention.

FIG. 3 is an explanatory diagram of the JPEG method.

FIG. 4 is a flowchart showing the operation of this embodiment.

FIG. 5 is an explanatory diagram showing the operation of the present embodiment.

FIG. 6 is an explanatory diagram of a spatial frequency spectrum of an original image of the present embodiment and image data obtained by subjecting the original image to compression / expansion processing.

Claims (4)

[Claims] 1. A method for evaluating compressed / decompressed image data, wherein a spatial frequency spectrum of image data of an original image before compression is obtained, and a spatial frequency spectrum of image data of image after compression / decompression processing is obtained. A method for evaluating compressed / decompressed image data, characterized in that the image data is evaluated by comparing the two spatial spectra. 2. A method for evaluating compressed / decompressed image data, wherein a spatial frequency spectrum of image data of an original image before compression is obtained, and a spatial frequency spectrum of image data of image after compression / decompression processing is obtained. The compressed / decompressed image data is characterized in that a spatial frequency spectrum in a frequency region that affects human visual characteristics is extracted from the two spatial frequency spectra, and the extracted spatial frequency spectra are compared. Evaluation method. 3. A first input means for inputting image data of an original image, a second input means for inputting compressed / decompressed image data, and an image input from the two input means. Spatial frequency spectrum calculation means for obtaining the spatial frequency spectrum of the data, spatial frequency spectrum comparison means for comparing the two spatial frequency spectrums, and evaluation of image deterioration based on the comparison result obtained by the spatial frequency spectrum comparison means. And an evaluation result output means for outputting an evaluation result, the evaluation device for the compressed / decompressed image data. 4. A first input means for inputting image data of an original image, a second input means for inputting compressed / decompressed image data, and an image input from the two input means. Spatial frequency spectrum calculation means for obtaining a spatial frequency spectrum of data, spatial frequency spectrum extraction means for extracting a spatial frequency spectrum of a predetermined frequency region from the two spatial frequency spectra, and the obtained 2
Spatial frequency spectrum comparing means for comparing spatial frequency spectra of two predetermined frequency regions, and evaluation result outputting means for evaluating image deterioration based on the comparison result obtained by the spatial frequency spectrum comparing means and outputting the evaluation result. An apparatus for evaluating compressed / decompressed image data, comprising: