JPH03133290A - Picture coder - Google Patents

Abstract

PURPOSE:To improve ease of observance of a decoded picture by providing an in-area information compression means compressing a feature extracted by an area extraction means and compressing split information of an partial area of an input picture and the feature in the local area separately. CONSTITUTION:A picture input section 211 converts a picture into a digital data and fetches the result. An area split section 211 divides the inputted picture into a local area whose feature is nearly uniform such as an area whose density is nearly uniform. The area division section 221 combines small areas with high similarity sequentially to form local areas and outputs the result of final area split to an area split information compression section 231, an area extraction section 241 and an in-area information compression section 251 as the area division information respectively, then to the area division information compression section 231, the result in matching with an average density in each area is outputted. The area division information compression section 231 extracts a contour line of the area based on the result of division obtained by the area division section 221 and compresses the result.

Description

[Detailed Description of the Invention] [Summary] Regarding an image encoding device that encodes digital image data, the input image is divided into partial regions according to characteristics, with the aim of improving the visibility of a restored image. a region dividing means for outputting information; a dividing information compressing means for compressing the dividing information; a region extracting means for extracting features corresponding to each partial region of the input image based on the dividing information; and intra-area information compression means for compressing the features of the partial area of the input image, and is configured to separately compress the division information of the partial area of the input image and the features within this partial area.

[Industrial application field]

The present invention relates to an image encoding device that encodes digital image data.

For example, in a system that provides images stored in a computer to a user via a low-speed communication line, efficient image compression techniques are required to reduce the time required to transmit image data. In addition, regarding image display, if each pixel is displayed at 1@, it will take a considerable amount of time to grasp the outline of one image, so it is possible to display hierarchical images that can quickly grasp the entire image. There is a need for image compression and decompression technology.

[Conventional technology]

FIG. 8 shows the configuration of a system for transmitting and receiving images stored in a computer. Camera 811 or scanner 8
The still image read in step 13 is compressed by a compression unit 823 in a computer 821, and this compressed data is stored in an image database 831.

When the user specifies the desired image, the image database 8
The compressed data in 31 is read out and sent to the user side via the communication line. On the user side, the compressed data is restored by the restoration unit 841, and the desired still image is displayed on the display device 851.

The communication capacity of the communication line through which compressed image data is sent and received in this way is generally 9.6Kbps or 64K.
bps, and the image information is enormous compared to this communication capacity, so even if high-efficiency encoding is performed, it will take several seconds to several tens of seconds to display one screen.

FIG. 9 shows a conventional image compression method. Figure (a) shows a method that creates pyramid-structured data from an original image, obtains images with multiple resolutions, and then compresses them. By creating pixels one layer higher, images corresponding to multiple layers are output.

Pyramid-structured data is obtained by compressing the images of each layer.

When transmitting image data, compressed data for images with a lower resolution (compressed data corresponding to an upper layer) is gradually transmitted, and compressed data for images with a higher resolution is gradually transmitted. On the receiving side, images of each resolution are displayed when the reception and restoration of the compressed data corresponding to each layer is completed.

In addition, in the same figure (b), the original image is divided into multiple small partial areas (
In this method, each block is divided into blocks) and orthogonal transformation is applied to each block. Examples of orthogonal transformation include Fourier transformation, Hadamard transformation, and discrete cosine transformation. Since an image generally has a high correlation between adjacent pixels, power is concentrated in low frequency components by performing orthogonal transformation. Therefore, by allocating a large number of bits to represent the low-frequency spectrum (-quantization) and representing the high-frequency spectrum by assigning a small number of bits, the overall data amount can be reduced and data compression can be achieved. conduct.

When transmitting image data, compressed data corresponding to a high frequency spectrum is sequentially transmitted starting with compressed data corresponding to a low frequency spectrum. On the receiving side, the image is restored by performing inverse transformation based on the received spectral components at appropriate intervals. This allows images with gradually higher resolution to be displayed.

FIG. 10 shows a display example when an image compressed by the two methods shown in FIG. 9 is restored on the receiving side. 10th
Figure (a) shows an image restored from a human image (original image) based on compressed data with a pyramid structure, and Figure (b) shows an image restored from an input image based on compressed data using orthogonal transformation. ing.

In FIG. 10(a), since a low resolution image is restored based on a small amount of information, the outline of the image can be grasped in a relatively short time. Furthermore, in FIG. 10(b), since the portion corresponding to the low frequency spectrum, that is, the rough appearance of the image, is restored first, the outline of the image can be grasped in a relatively short time.

[Problem to be solved by the invention]

By the way, with the two image compression methods mentioned above, it is possible to grasp the image in a relatively short time, but because the resolution of the displayed image is gradually increased, the outline of the image is There is a problem in that the restored image is difficult to see because parts become angular, step-like, or blurred.

In particular, when humans view images, it is known that contour information in the images plays an important role.
If the contour information is stored at high resolution, the image appears to be of fairly good quality to the human eye even if the information within the contour is low resolution. Therefore, by utilizing such properties, it is possible to improve the visibility of the restored image.

The present invention was created in view of these points, and an object of the present invention is to provide an image encoding device that improves the visibility of restored images.

[Means to solve the problem]

FIG. 1 is a principle block diagram of an image encoding device according to the present invention.

In the figure, region dividing means 111 divides an input image into partial regions according to characteristics and outputs division information.

The division information compression means 121 compresses the division information.

The region extracting means 131 extracts features corresponding to each partial region of the human image based on the division information.

The intra-region information compression means 141 compresses the features extracted by the region extraction means 131.

Therefore, the overall configuration is such that the division information of each partial area into which the input image is divided and the features within this partial area are separately compressed.

[For production]

When an image is input, the region dividing means 111 divides the input image into partial regions according to the characteristics and outputs division information, and this division information is compressed by the division information compression means 121.

Further, the region extraction means 131 extracts features corresponding to each partial region of the input image based on the division information output from the region division means 111, and the features of each partial region are compressed by the intra-region information compression means 141. be done.

In the present invention, by separately compressing the division information of each partial area into which an input image is divided and the features within this partial area, the divided parts (for example, the outline) of the image can be compressed when restoring the compressed data. It will be possible to restore it first.

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail based on the drawings.

FIG. 2 shows the configuration of an embodiment of the present invention.

Here, the correspondence relationship between the embodiment of the present invention and FIG. 1 is shown (.

The region dividing means 111 corresponds to the region dividing section 221.

The division information compression unit 121 includes the area division information compression unit 231
corresponds to

The region extraction means 131 corresponds to the region extraction section 241.

Intra-area information compression means 141. is the area information compression unit 25
Corresponds to 1.

Examples of the present invention will be described below assuming that the correspondence relationship as described above exists.

In FIG. 2, the image encoding device 210 includes an image input section 211. Area dividing section 221. Region division information compression unit 231
．． eNN oil extraction 241. An intra-region information compression unit 251 is provided.

The image input unit 211 converts an image into digital data and inputs the digital data. The captured image is divided into area division section 2
21 and the area extraction unit 241.

The region dividing unit 221 divides the input image into partial regions having substantially uniform features. For example, it is divided into regions with substantially uniform density. As a specific processing method, for example, "Overview of Image Processing Algorithms (5) -? il area division-,
Yokoya, Tomita, Tamura, Report of Electronic Technology Research Institute, Vol. 44, No. 78, pp. 504-520, 1981 J. The details include the spirit merge method (separation and integration method), one pixel combination method, etc. There is.

The area dividing unit 221 sequentially combines small areas with high similarity to form partial areas, and uses the final area dividing result as area dividing information to the area dividing information compression unit 231. ? iJI
It is output to each of the area extraction section 241 and the intra-area information compression section 251. Furthermore, the average density value within each region is also output to the region division information compression unit 231.

FIG. 3 shows an example of area division information. As shown in the figure,
For example, the input image is divided into five partial regions (each region Nα is "1").
~ "5"), the area Nα that includes the pixel of interest is the area division information corresponding to each pixel.
is output.

The region division information compression unit 2 31 extracts and compresses the contour line of the region based on the division result (region division information) obtained by the region division unit 221.
33 and a chain encoding section 235.

The contour line extraction unit 233 traces and extracts the contour line around the outer periphery of the partial area based on the area division information for each pixel inputted from the area division unit 221, and the chain encoding unit 235 traces and extracts the contour line around the outer circumference of the partial area. Compress the contour using chain code. The chain code output from the chain encoder 235 is output from the image encoder 210 together with the average density value in each divided area, and is sent to the image database 261.
is accumulated in FIG. 4 shows an outline of the chain code. Figure (a) is an example of a chain code, which basically consists of eight types of direction codes (
(see figure (b))) to express the contour line.

Further, the region extraction section 241 extracts each partial region from the input image, which is the original image, based on the region division information output from the region division section 221, and converts the density data of each pixel in each extracted partial region into the region. Output as internal information.

The intra-area information compression unit 251 compresses the intra-area information output from the area extraction unit 241 by applying orthogonal transform encoding, and includes a block dividing unit 253 and an orthogonal transform unit 255.
9 layering section 257. It consists of a quantization section 259.

The block dividing unit 253 divides each partial area divided by the area dividing unit 221 into blocks of a predetermined size, for example, 8 layers and 8 pixels.

The orthogonal transformation unit 255 performs orthogonal transformation on each block divided by the block division unit 253.

In this orthogonal transformation, a two-dimensional orthogonal transformation is applied to each block, but if the block includes a contour line, pixels outside the contour line are interpolated with pixels similar to pixels inside the contour line ( For example, by replacing the pixel with a pixel having an average pixel value inside the contour), and performing orthogonal transformation by equalizing the information using the information inside the contour.

By interpolating pixels outside the contour line with pixels inside the contour line in this way, orthogonal transformation can be performed in an optimal state, making it possible to efficiently compress information.

Figure 5 shows an example of interpolation of pixels outside the contour line,
A case is shown in which pyramid structure data is created using four adjacent pixels as a unit. For example, the average value of the 0 layer (each block extracted from the input image) excluding non-target pixels is taken as the pixel value one layer higher, and the pixel value one layer higher is, for example, 0
This is the value of the pixel outside the layer area.

The hierarchization unit 257 hierarchizes the spectral components subjected to the two-dimensional orthogonal transformation in the orthogonal transform unit 255 in accordance with the frequency, and an example of hierarchization is shown in FIG.

■, ■, etc. in the figure indicate each hierarchical layer, and "
0'' indicates orthogonal transformation data represented by bit ``o''.

The quantization unit 259 expresses each spectral component hierarchized by the hierarchization unit 257 with the number of bits corresponding to its size, and the amount of data for expressing the entire input image is reduced. The compressed data obtained in this way is output in order from the low frequency component and stored in the image database 261.

A case in which the compressed data thus accumulated in the image database 261 is transmitted via the communication line shown in FIG. 8, for example, will be described below.

The image search unit 271 searches the image database 261 in response to a search request from a user and reads compressed data of the corresponding image. In transmitting this compressed data, first the compressed data of the region division information, that is, the chain code of the outline information of each partial region and the average density value within the partial region are transmitted. On the receiving side, the initial image is displayed when the reception and restoration of this information is completed.

Next, the compressed data of the area information is transmitted, and high frequency spectral components are transmitted sequentially from low frequency spectral components. On the receiving side, the image is restored and displayed based on the received data at appropriate intervals. This makes it possible to gradually increase the resolution within the area according to the amount of received data.

FIG. 7 shows an outline of the process of image restoration on the receiving side. First, since the image is restored based on the contour information of the partial area and the average density value within the partial area, a screen is displayed in which the inside of each partial area is expressed using the average density value. Since the outline is clear on this screen, the outline of the image can be grasped and it is easy for the user to see. next,
Since the image within each region is restored based on the low-frequency spectral component and the high-frequency spectral component of the intra-region information, the resolution increases sequentially and the original image is finally restored.

In the embodiment of the present invention, chain encoding and orthogonal transform encoding are combined, but any encoding method may be used in combination, as long as the contour portion of the original image can be encoded separately.

In addition, although density is used as an example of a feature, other characteristics may be used as long as it is a feature that expresses each pixel of the input image (for example, if each pixel is expressed by a luminance signal and a color signal, luminance information and color information).

Further, in the embodiment, a case has been described in which the compressed data stored in the image database 261 is transmitted via a communication line, but it may be directly displayed on a display device provided in a computer system or the like.

〔Effect of the invention〕

As described above, according to the present invention, by separately compressing the division information of each partial region into which an input image is divided and the features within this partial region, the divided portions ( For example, this method is extremely useful in practical terms because it can improve the visibility of the restored image by restoring the contour of the image first.

[Brief explanation of the drawing]

Fig. 1 is a principle block diagram of the image encoding device of the present invention, Fig. 2 is a block diagram of an embodiment of the present invention, Fig. 3 is an explanatory diagram of area division information, and Fig. 4 is an explanatory diagram of chain code. , Fig. 5 is an explanatory diagram of interpolation of pixels outside the contour line, Fig. 6 is an explanatory diagram of hierarchization of orthogonal transformation data, Fig. 7 is an explanatory diagram of the restored image of the example, and Fig. 8 is an illustration of image communication. A system configuration diagram, FIG. 9 is an explanatory diagram of a conventional image compression method, and FIG. 10 is an explanatory diagram of a restored image in the conventional example. In the figure, 111 is a region division means, 121 is a division information compression means, 131 is a region extraction means, 141 is an intra-region information compression means, 210 is an image encoding device, 211 is an image input section, 221 is a region division section, 231 233 is a contour extraction unit, 235 is a chain encoding unit, 241 is a region extraction unit, 251 is an intra-region information compression unit, 253 is a block division unit, 255 is an orthogonal transformation unit, 257 is a layering unit 259 is a quantization unit, 261 is an image database, and 271 is an image search unit. Input image area division information Explanation diagram of area division information Figure 3 Chain code (0760021...) (a) 8-direction code (b) Explanation diagram of interpolation of pixels outside the contour line Figure 5 High frequency orthogonal transformation data Explanatory diagram of layering Go to Figure 6, page 21 ↑ ↑ ↑ ↑

Claims (1)

[Claims] (1) a region dividing means (111) that divides an input image into partial regions according to characteristics and outputs division information; a division information compression means (121) that compresses the division information; and a division information compression means (121) that compresses the division information. , region extraction means (131) for extracting features corresponding to each partial region of the input image, and intra-region information compression means (141) for compressing the features extracted by the region extraction means (131). . An image encoding apparatus, characterized in that the image encoding apparatus is configured to separately compress division information of a partial area of the input image and features within this partial area.