JPH0514737A - Picture processor - Google Patents

Abstract

PURPOSE:To improve compression ratio while suppressing the deterioration of picture quality. CONSTITUTION:An input picture is DCT-processed by a DCT processing circuit 13 every block of 8X8-picture elements, and it is discriminated whether the input picture is a continuously varying picture like a photograph or a binary picture like a character and a pattern, etc., by a picture discrimination circuit 14 on the basis of an obtained conversion coefficient. As for an area discriminated to be the continuously varying picture, the area of 16X16-picture elements is made one block of 8X8-picture elements by converting resolution by a resolution conversion circuit 15, and the input picture is DCT-processed again by the DCT processing circuit 13, and the conversion coefficient 31 is sent to a quantization circuit 16. As for the area discriminated to be the binary picture, the conversion coefficient 31 is sent to the quantization circuit 16 while the resolution is not converted. The conversion coefficient is quantized by the quantization circuit 16, and is encoded by an encoding circuit 17, and compressed data is obtained. Data in which the compressed data and the discriminated result of the picture are coordinated with each other is generated by a control data generation circuit 18.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus for compressing and expanding multi-valued images and color still images.

[0002]

2. Description of the Related Art In recent years, image information has been widely used and image processing technology has been developed. One of the image processings is compression and expansion processing. Conventionally, as a method of compressing a multi-valued image or a color still image, there is a method of using orthogonal transform such as discrete cosine transform (hereinafter referred to as DCT).

FIG. 9 is a block diagram showing a conventional image processing apparatus for compressing an image by DCT processing. This image processing device includes a two-dimensional DCT processing circuit 52 that performs DCT processing for each block of a predetermined number of pixels of an input image 51, and a quantization circuit that quantizes a transform coefficient obtained by the two-dimensional DCT processing circuit 52. 53 and an encoding circuit 54 for encoding the data quantized by the quantization circuit 53. In this image processing device, the two-dimensional DCT processing circuit 52 divides an image into blocks, for example, in units of 8 × 8 pixels, and performs DCT processing for each block. Then, the obtained transform coefficient is quantized by the quantizing circuit 53 on the basis of the human visual characteristic that the sensitivity is high for the high frequency components of the image and high for the low frequency components, and The coefficient corresponding to the high frequency component is cut so that the coefficient corresponding to the low frequency component remains. Then, the quantized data is encoded by the encoding circuit 54 to obtain the compressed image data 55.

[0004]

In the conventional compression method, DCT processing is performed for each block of 8 × 8 pixels, and therefore it is necessary to further cut the transform coefficient corresponding to the high frequency component in order to perform higher compression. There is. However,
There is a problem in that the image quality is deteriorated when the image that has undergone higher compression is expanded.

Therefore, an object of the present invention is to provide an image processing apparatus capable of increasing the compression rate while suppressing the deterioration of the image quality.

According to another aspect of the invention, there is provided an image processing apparatus, which comprises an orthogonal transform means for orthogonally transforming an image for each block having a predetermined number of pixels, and a pixel within the region for each region having a predetermined size of an input image. Image discrimination means for discriminating the characteristics of the density distribution of
The resolution in the area is switched according to the discrimination result of the image discrimination means to switch the number of pixels in the area, and the resolution switching means for switching the size of the input image corresponding to the block and the orthogonal transformation by the orthogonal transformation means are obtained. It is provided with a quantizing means for quantizing the coefficient and an encoding means for coding the data quantized by the quantizing means.

In this image processing apparatus, the image discriminating means discriminates the characteristics of the density distribution of the pixels in each area of a predetermined size of the input image, and the resolution switching means responds to the discrimination result. , The resolution in the area is switched, and the number of pixels in the area is switched. Then, the orthogonal transformation is performed by the orthogonal transformation unit for each block having a predetermined number of pixels of the image after the resolution is switched,
The obtained coefficient is quantized by the quantizing means and coded by the coding means.

The image processing apparatus according to the second aspect of the present invention is, for each predetermined region of the input image, an image discriminating means for discriminating whether or not the spatial change in the density of the pixels in the region is continuous. With respect to the area determined to be continuous by the image determination means, the resolution conversion means for converting the resolution to reduce the number of pixels in the area, and the area determined not to be continuous by the image determination means The image is orthogonally transformed for each block having a predetermined number of pixels of the input image, and for the area determined to be continuous by the image determining means, for each block having a predetermined number of pixels of the image converted by the resolution converting means. Orthogonal transformation means for orthogonally transforming the image into
It is provided with a quantizing means for quantizing the coefficient obtained by the orthogonal transformation by the orthogonal transforming means, and an encoding means for encoding the data quantized by the quantizing means.

In this image processing apparatus, the image discrimination means discriminates, for each area of a predetermined size of the input image, whether or not the spatial change in the density of the pixels in the area is continuous. For the region determined to be continuous, the resolution is converted by the resolution conversion unit to reduce the number of pixels in the region, and then the orthogonal conversion unit performs orthogonal conversion for each block having a predetermined number of pixels. On the other hand, with respect to the area determined not to be continuous, orthogonal transformation is performed by the orthogonal transformation unit for each block having a predetermined number of pixels of the input image. The coefficient obtained by the orthogonal transform is quantized by the quantizing means and coded by the coding means.

An image processing apparatus according to a third aspect of the present invention is the image processing apparatus according to the first or second aspect of the present invention, further comprising data quantized by the quantization means for each block subjected to orthogonal transformation by the orthogonal transformation means. A means for associating with the discrimination result of the image discriminating means is provided.

[0011]

Embodiments of the present invention will be described below with reference to the drawings. 1 to 8 relate to an embodiment of the present invention.

FIG. 1 is a block diagram showing the arrangement of the image processing apparatus of this embodiment. The image processing apparatus according to the present embodiment includes an image scanner 11 as an image input unit, and an image memory 12 that stores an image input by the image scanner 11. A two-dimensional DCT processing circuit 13 and a resolution conversion circuit 15 are connected to the image memory 12, and an image discrimination circuit 14 is connected to the two-dimensional DCT processing circuit 13 and the resolution conversion circuit 15. The two-dimensional DCT processing circuit 13 sets D for each block having a predetermined number of pixels.
CT processing is performed. The image discriminating circuit 14 calculates the feature amount of the image from the conversion coefficient 31 obtained by the two-dimensional DCT processing circuit 13 for each predetermined region of the input image, and the region has a spatial density of pixels like a photograph. Whether it is a multi-valued image (hereinafter referred to as a continuously changing image) that continuously changes in a continuous manner, or a binary image such as a character or a figure (hereinafter referred to as a binary image). Is to determine. The resolution conversion circuit 15 uses the discrimination signal 3 from the image discrimination circuit 14.
In accordance with 2, the resolution of the area determined by the image determination circuit 14 to be a continuously changing image is converted to reduce the number of pixels. The image 33 whose resolution is converted by the resolution conversion circuit 15 is the two-dimensional DCT processing circuit 13
And is subjected to DCT processing. The two-dimensional DCT processing circuit 13 orthogonally transforms, in accordance with the discrimination signal 32 from the image discrimination circuit 14, an orthogonal transformation for each block having a predetermined number of pixels of the input image for the region discriminated as a binary image. 31 is output, and with respect to the area determined to be the continuously changing image, the conversion coefficient 31 that is orthogonally transformed for each block of the predetermined number of pixels of the image 33 whose resolution is converted by the resolution converting circuit 15 is output.

The image processing apparatus further includes a quantizing circuit 16 for quantizing the transform coefficient from the two-dimensional DCT processing circuit 13.
And an encoding circuit 17 for encoding the data quantized by the quantization circuit 16 and outputting the compressed data.
And a management data creation circuit 18 for creating management data in which the compressed data encoded by the encoding circuit 17 and the header data indicating the determination result of the image determination circuit 14 are associated with each other. The management data created by the management data creation circuit 18 is stored in the hard disk 19
Stored in or sent to.

Further, as shown in FIG. 1, the apparatus for decompressing the image data compressed by the image processing apparatus having the above-described configuration is such that the management data is divided into compressed data and header data. A separation circuit 21, a decompression circuit 22 for decompressing the compressed data separated by the management data separation circuit 21, and a management data separation circuit 21.
An interpolation circuit 23 that performs an interpolation process on the output data of the decompression circuit 22 in the area for which the resolution conversion has been performed based on the header data separated by
3 and an image memory 24 for storing the output image of FIG. The decompression circuit 22 includes a circuit that decodes the encoded compressed data, a circuit that restores the decoded data into transform coefficients, and a circuit that performs inverse DCT processing on the transform coefficients to obtain pixel data.

Next, the operation of the image processing apparatus of this embodiment will be described with reference to FIGS.

FIG. 2 is a flow chart showing the operation of the image processing apparatus of this embodiment. The image processing apparatus first executes the image scanner 1 in step (hereinafter referred to as S) 101.
The image is input by 1 and stored in the image memory 12.
This image is, for example, a multi-valued image or a color still image in which a binary image area such as characters and figures and a multi-valued image area such as a photograph are mixed, and here, the density of one pixel is 8 bits. Let it be a multi-valued image represented by.

Next, in step S102, the image stored in the image memory 12 is 8 × by the two-dimensional DCT processing circuit 13.
DCT processing is performed for each block of 8 pixels. FIG. 3 shows one block of 8 × 8 pixels, and the numerical values in the figure show an example of the density for each pixel. Further, FIG. 4 shows an example of transform coefficients obtained by subjecting this block to DCT processing. Note that the numerical values in FIG. 4 conceptually represent the conversion coefficient, and are not the values actually processed. Further, the transform coefficient shown in FIG. 4 is shown as normalized so that the range of the transform coefficient obtained by the DCT process is 0 to 255 (8 bits).

Next, in step S103, the transform coefficient of the 8 × 8 pixel block obtained by the two-dimensional DCT processing circuit 13 is input to the image discriminating circuit 14, and the feature amount of the image is calculated from this transform coefficient. Next, in step S104, it is determined whether the image of the 8 × 8 pixel block is a continuously changing image or a binary image based on the feature amount obtained in step S103. Here, as the feature amount, for example, the value of the conversion coefficient corresponding to one or more predetermined frequency ranges is used. Continuous change images and characters such as photographs
The distribution (spectrum) of the values of the conversion coefficient is different from that of a binary image such as a figure. Therefore, a standard feature amount of the continuously changing image and the binary image is statistically obtained in advance from a plurality of continuously changing images and the binary image, and the standard feature amount and the 8 × 8 pixel block By comparing the image with the feature amount of the image, it is possible to determine whether the image of the 8 × 8 pixel block is a continuously changing image or a binary image. In the present embodiment, if the feature amount of the 8 × 8 pixel block is equal to or less than a certain value determined from the standard feature amount of the continuously changing image and the binary image, it is determined that the image is the continuously changing image. If not, it is determined that there is a binary image.

In this embodiment, after determining whether the image of each block is a continuously changing image or a binary image for the four blocks, the area formed by the four blocks is a continuously changing image or a binary image. It is judged whether it is an image and the subsequent processing is different. FIG. 5 shows the order of blocks for performing DCT processing and feature amount calculation. 2x2 shown in this figure
For the block region 40, it is determined whether each of the four blocks is a continuous change image or a binary image, and if, for example, two or more blocks are determined to be continuous change images,
The region 40 is determined to be a continuously changing image, and in other cases, the region 40 is determined to be a binary image.

When it is determined in S104 that the area 40 is a continuously changing image, the resolution conversion circuit 15 performs resolution conversion of the area 40 in response to the determination signal 32 from the image determination circuit 14 in S105. . That is, the resolution conversion circuit 15 reads the density of each pixel in the area 40 from the image memory 12, obtains the average of the densities of four adjacent pixels, and sets this as the density of one new pixel. As a result, the area 40 of 16 × 16 pixels is 8 ×
It is one block of 8 pixels.

The image 33 of the new block created by this resolution conversion is input to the two-dimensional DCT processing circuit 13 for DCT processing in S106, and the obtained transform coefficient 31 is sent to the quantization circuit 16. To be

On the other hand, if it is determined in S104 that the region 40 is a binary image, resolution conversion is not performed for the region 40 and the conversion coefficient 31 of each block is not converted.
Are sent from the two-dimensional DCT processing circuit 13 to the quantization circuit 16. The two-dimensional DCT processing circuit 13 has a memory that temporarily holds the transform coefficient 31 of each block of the area 40, and according to the discrimination signal 32 from the image discrimination circuit 14,
When the resolution conversion is not performed, the conversion coefficient of each block stored in the memory is output, and when the resolution conversion is performed, the conversion coefficient 31 obtained by DCT processing the image 33 from the resolution conversion circuit 15 is output.

The transform coefficient sent to the quantization circuit 16 is S
At 107, the quantization circuit 16 quantizes. This quantization cuts the coefficient corresponding to the high frequency component of the image, based on the human visual characteristic that it has low sensitivity to high frequency components of the image and high sensitivity to low frequency components,
The coefficient corresponding to the low frequency component remains. For example,
The data length is changed between the low frequency side and the high frequency side such that the coefficient corresponding to the low frequency component is quantized by 8 bits and the coefficient corresponding to the high frequency component is quantized by 4 bits. Alternatively, the amount of information may be reduced by decreasing the actual coefficient value on the higher frequency side. The amount of information is reduced by such quantization.

The data quantized by the quantization circuit 16 is encoded by the encoding circuit 17 in S108 and is input to the management data creation circuit 18 as compressed data. As this encoding, two-dimensional run-length encoding standardized by "Standardization Committee for Coding of Color Still Images (Joint Photographic Experts Group)" can be used.

The management data creation circuit 18 uses S109.
Then, the compressed data from the encoding circuit 17 and the image discrimination circuit 1
The management data is created by using the discrimination signal 32 from the No. 4.

FIG. 6 is an explanatory diagram showing the structure of management data. This management data has a structure in which header data is added to the beginning of compressed data for each processing unit. The header data includes information as to whether compression processing was performed with 8 × 8 pixels of the input image as one block or 16 × 16 pixels of the input image as one block. When 8 × 8 pixels are one block, the compressed data of the block of 8 × 8 pixels of the input image comes after the header data. 1
When 6 × 16 pixels are set as one block, the compressed data of the block in which four blocks of 8 × 8 pixels of the input image are converted into one to be one comes.

FIG. 7 is an explanatory diagram showing another example of the structure of the management data. This management data has a structure in which one header data is added to the head of the compressed data for one page. FIG. 8 shows the structure of header data in the case of an image of 200 × 300 blocks. This header data has 1-bit data for each block of 8 × 8 pixels of the input image. For example, if the block is a continuously changing image, the bit of the header data corresponding to the block is "1", and if it is a binary image, the bit of the header data corresponding to the block is "0". , Input image 8x
It is possible to distinguish whether or not a plurality of blocks of 8 pixels are collectively processed. By managing the compressed data as shown in FIGS. 6 to 8, it is possible to easily restore the image in the decompression process.

The management data created in this way is converted into S
At 110, the data is stored in the hard disk 19 or transmitted, and the process ends.

When decompressing the compressed data, as shown in FIG. 1, the management data separating circuit 21 separates the compressed data and the header data, and the decompressing circuit 22 expands the compressed data. Restore pixel data. This pixel data is input to the interpolation circuit 23. This interpolation circuit 23
Performs an interpolation process on the output data of the decompression circuit 22 for the area for which the resolution conversion has been performed, based on the header data separated by the management data separation circuit 21.
Data of 16 × 16 pixels is restored from data of 8 × 8 pixels. Since a known method can be used as the interpolation processing, the description thereof will be omitted. Further, for the area where the resolution conversion has not been performed, the output data of the decompression circuit 22 is output as it is without performing the interpolation processing. The output data of the interpolation circuit 23 is stored in the image memory 24, and the image for each page is restored.

As described above, according to this embodiment, 8
Assuming that each block of 4 blocks has × 8 pixels as one block, that region is a continuously changing image or a binary image, and resolution conversion is not performed for the region that is determined to be a binary image. The DCT processing is performed for each block of 8 × 8 pixels, and the area determined to be the continuous change image is
After the resolution is converted and the number of pixels is reduced so that 16 × 16 pixels become one block of 8 × 8 pixels, DCT processing is performed. This is based on the fact that it is better for a continuous-change image such as a photograph to be able to preserve the gradation than the resolution, and for a binary image such as a character or a figure to be able to preserve the resolution than the gradation. ing. That is, the image quality of the continuously changing image is not deteriorated even if the resolution is coarsened. Therefore, the compression rate can be increased while suppressing the deterioration of the image quality by roughening the resolution and processing the four blocks collectively only for the region of the continuously changing image.

The present invention is not limited to the above embodiment,
For example, the present invention can be applied to an image processing apparatus that performs compression processing using other orthogonal transformation such as Fourier transformation and Hadamard transformation.

[0032]

As described above, according to the invention described in claim 1 or 2, the resolution in the area is switched according to the characteristics of the density distribution of the pixels in the predetermined area, or the space of the density of the pixels is changed. The resolution in the area is switched according to whether the change is continuous or not, and then the image is orthogonally transformed for each block of a predetermined number of pixels, so that the compression rate can be reduced while suppressing the deterioration of the image quality. The effect is that it can be increased.

According to the third aspect of the invention, since the quantized data and the image discrimination result are made to correspond to each other for each block subjected to the orthogonal transformation by the orthogonal transformation means, the decompression processing is performed. It has the effect of making it easier.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an image processing apparatus according to an exemplary embodiment of the present invention.

FIG. 2 is a flowchart showing an operation of the image processing apparatus of the embodiment.

FIG. 3 is an explanatory diagram showing one block of 8 × 8 pixels.

FIG. 4 is an explanatory diagram showing an example of transform coefficients obtained by performing DCT processing on the block of FIG.

FIG. 5 is an explanatory diagram showing the order of blocks for performing DCT processing and feature amount calculation in the image processing apparatus according to the embodiment.

FIG. 6 is an explanatory diagram showing a structure of management data in the image processing apparatus of the embodiment.

FIG. 7 is an explanatory diagram showing another example of the structure of management data in the image processing apparatus of the embodiment.

8 is an explanatory diagram showing a structure of header data in FIG. 7. FIG.

FIG. 9 is a block diagram showing a configuration of a conventional image processing apparatus.

[Explanation of symbols]

13 ... Two-dimensional DCT processing circuit, 14 ... Image discrimination circuit, 1
5 ... Resolution conversion circuit, 16 ... Quantization circuit, 17 ... Encoding circuit, 18 ... Management data creation circuit

Claims (3)

[Claims] 1. An orthogonal transformation means for orthogonally transforming an image for each block of a predetermined number of pixels, and an image discrimination for discriminating a characteristic of a density distribution of pixels in the region for each region of a predetermined size of an input image. Means, resolution switching means for switching the resolution in the area according to the discrimination result of the image discrimination means to switch the number of pixels in the area, and for switching the size of the input image corresponding to the block, and the orthogonal transformation. An image processing apparatus comprising: a quantizing means for quantizing a coefficient obtained by orthogonal transformation by means; and an encoding means for encoding the data quantized by the quantizing means. 2. An image discriminating means for discriminating whether or not a spatial change in the density of pixels in the predetermined region of the input image is continuous, and the image discriminating means is arranged so as to be continuous. For a region determined to be present, resolution conversion means for converting the resolution to reduce the number of pixels in the region, and for a region determined not to be continuous by the image determination means, a block having a predetermined number of pixels in the input image. The image is orthogonally transformed for each of the regions, and the region that is determined to be continuous by the image determination unit is orthogonally transformed for each block having a predetermined number of pixels of the image after being converted by the resolution conversion unit. Transforming means, quantizing means for quantizing the coefficients obtained by the orthogonal transforming by the orthogonal transforming means, and coding means for encoding the data quantized by the quantizing means. An image processing apparatus comprising: 3. The means for associating the data quantized by the quantizing means with the discrimination result of the image discriminating means for each block subjected to the orthogonal transformation by the orthogonal transforming means. The image processing device according to claim 1.