JPH09161078A - Method for device for encoding, and method and device for decoding - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the efficiency of encoding by encoding only information included within the outline of a texture closed area. SOLUTION: A picture including a texture-closed area and an edgeclosed area is inputted to an encoding device 100 as an input picture signal DD0. A luminance information encoding circuit DP1 finds out a local luminance component indicating the general luminance information of the picture from the signal DD0 inputted to the device 100, encodes the component, supplies the encoded local luminance component DD1 to an edge picture reconstituting processing circuit DP3, and outputs component DD1 as luminanceencoded data ED1. On the other hand, an outline encoding circuit DP2 extracts the outline part of the picture from the signal DD0 by an edge area extraction circuit and encodes the extracted information. The circuit DP2 supplies the encoded outline part information DD2 to the circuit DP3 and outputs the information DD2 as outline encoded data ED2.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a closed area coding method and coding apparatus using a high efficiency coding method, which is characterized in that the contour lines of an object in an image are preserved with emphasis. The present invention relates to a coding method or a decoding method and a decoding device for a code of a closed region obtained by the above coding device.

[0002]

2. Description of the Related Art For example, in a transmission device for transmitting an image by a transmission medium having a limited transmission capacity, or a recording / reproducing device for recording an image on a recording medium and reproducing the image from the recording medium, An encoding device to which a high efficiency encoding method is applied is used.

The high-efficiency coding method for image signals is a coding method for avoiding deterioration of image quality as much as possible and reducing redundancy by utilizing the high correlation of image signals. It is indispensable when transmitting or recording an image on a recording medium.

As a high-efficiency coding system for image signals, a predictive coding system and a discrete cosine transform (DCT) are used.
Orthogonal transform coding method represented by osine transform),
And sub-band coding methods such as wavelet transform.

Intraframe DP is used as a predictive coding method.
There is CM (Differential Pulse Code Modulation) and the like, and the predictive coding method uses spatial or temporal correlation to handle an image in pixel units, and an original pixel and a neighboring pixel of the decoded original pixel. This is a method of encoding the difference between and. In such a predictive coding system, the required compression rate is 1 /
This method is effective when the compression rate is not so high as about 2 to 1/4, but is not suitable when a higher compression rate is required.

On the other hand, the orthogonal transform coding method and the sub-band coding method are used when the compression rate is high such that the required compression rate is 1/10 or more.
A coding method using CT is generally used. This is because the DCT has a high-speed algorithm and can be easily implemented in hardware. Also, DCT is
It is also used in JPEG, MPEG, and other international standards.

The basic principle of the above-mentioned coding system using DCT (hereinafter referred to as DCT coding system) is obtained by DCT by utilizing the characteristic that the power of the low frequency component of the image signal is extremely large. When quantizing the frequency component of the obtained image signal, the quantization step size of the low frequency component is made small and the quantization step size of the high frequency component is made large, so that the information amount as a whole is compressed. It means that you can

However, the DCT coding method has a problem that block distortion and mosquito noise occur due to quantization, and in particular, it is caused by the coding processing in units of macroblocks. Block distortion becomes significant when the encoding speed is low.

Therefore, as an encoding method for the purpose of transmitting an image at an ultra-low bit rate and recording an image on a recording medium, human visual characteristics are particularly sensitive to the contour line of an object. In consideration of the above, there has been proposed an encoding method that realizes visually excellent restoration even at an extremely low bit rate by preserving the contour line portion in the original image.

As an encoding method for preserving the contour portion in the original image as described above, an edge reconstructed image reconstructed from the contour portion in the original image and pixels on both sides of the contour portion, Reproduce small changes in the original image by encoding the image and the texture image that is the difference image of the edge reconstructed image separately and adding the texture image to the edge reconstructed image when reconstructing There is an encoding method designed to do so.

FIG. 16 is a block diagram showing the structure of a texture image coding circuit in the coding apparatus to which the above coding method is applied. The texture image encoding process will be described below with reference to FIG.

First, an edge area signal BD0 obtained by extracting an edge area such as a contour of an object or a boundary line between objects in an original image is supplied to an edge area expanding circuit BP3 via an edge area buffer BP0. It Further, an edge reconstructed image signal (hereinafter, referred to as an edge reconstructed image signal) BD1 reconstructed from the extracted edge region and pixels on both sides of the edge region, and an original image signal (hereinafter, referred to as an edge reconstructed image signal). BD2 is a difference circuit B via an edge reconstructed image buffer BP1 and an original image buffer BP2.
Each is supplied to P4.

The edge area expansion circuit BP3 receives the edge area signal BD0 from the edge area buffer BP0,
An expansion edge area is obtained by using pixels in a certain range from an arbitrary pixel as an expansion mask area, and the obtained expansion edge area signal (hereinafter, expansion edge area signal) BD3 is supplied to the texture area separation circuit BP5.

On the other hand, the difference circuit BP4 has an edge reconstructed image signal BD1 from the edge reconstructed image buffer BP1.
And the original image signal BD2 from the original image buffer BP2, and the difference signal is used as the texture image signal BD4.
Is supplied to the texture area separation circuit P5.

The texture area separation circuit BP5 removes the expanded edge closed area MA from the texture image signal BD4 from the difference circuit BP4 as shown in FIG. 17 by using the expanded edge area signal BD3 from the edge area expansion circuit BP3. The macroblock generation circuit BP that extracts only the texture image signal of the texture area TA that is the range
6

The macroblock generation circuit BP6 receives the texture area signal BD from the texture area separation circuit BP5.
5, the texture area TA as shown in FIG.
8 × 8 pixel block processing is performed, and the texture area block signal DB6 obtained by the block processing is D
It is supplied to the CT conversion circuit BP7.

The DCT conversion circuit BP7 is provided with a texture area block signal DB from the macroblock generation circuit BP6.
6 is subjected to orthogonal transform processing in macro block units,
The orthogonal transform coefficient signal BD7 obtained by the orthogonal transform process is supplied to the quantization circuit BP8.

The orthogonal transform coefficient signal BD7 obtained by the DCT transform circuit BP7 becomes a quantized signal BD8 by the quantizer circuit BP8, and then the variable length coding circuit BP9.
And is output as a texture encoded signal BD9 via the output buffer BP10.

[0019]

However, when the texture area is coded by the conventional coding apparatus as described above, the macro block of the fixed size as shown in FIG. 18 is used. When all the pixels inside are coded, a macroblock that includes a texture region even in part has been the target of the coding process as a texture region block. Therefore, the actually encoded range is wider than the original texture area, and the encoding amount is increased.

Therefore, the present invention has been made in view of the above-mentioned conventional circumstances, and has the following objects.

That is, it is an object of the present invention to provide a coding method and a coding device that improve coding efficiency by coding only information within the contour of a texture closed region.

It is another object of the present invention to provide a coding method and a coding device which can further improve the coding efficiency by being adaptable to the motion compensation processing of the texture closed region for the moving image. .

Another object of the present invention is to provide a decoding method and a decoding apparatus for completely restoring the information coded by the above coding method or the above coding apparatus.

[0024]

In order to solve the above-mentioned problems, a coding method according to the present invention is a contour coded by extracting contour lines of a plurality of closed regions of indefinite shape existing in an image. A coding method for outputting line-coded data, wherein the contour lines of a plurality of closed regions of indefinite shape are used as pixel value distribution inside the closed region and the contour line shape of the closed region as a feature amount of each closed region. Extract. Then, each extracted contour line is encoded.

Further, the encoding method according to the present invention binarizes pixels inside the closed area and pixels outside the closed area. Then, the contour line is extracted by using the boundary pixel where the value of the binarized pixel changes as the contour pixel of the closed region.

Further, the coding method according to the present invention is characterized in that coding is performed for each contour line of each closed region.

Further, in the encoding method according to the present invention, an arbitrary point on the contour line is used as a start point, and the pixels on the contour line are sequentially traced from the pixel existing at the start point to perform the encoding.
Then, the encoding of the contour line is terminated at the time point of returning to the start point.

Further, the encoding method according to the present invention detects pixels on the contour line other than the target pixel existing in the window area centered on the target pixel on the contour line as candidate pixels. Then, an arbitrary candidate pixel among the detected candidate pixels is selected as a target pixel to be encoded next.

In addition, the encoding method according to the present invention is such that the pixels to be encoded next are arranged so that the pixels on both sides of the target pixel in the encoding direction of the pixels on the contour line always belong to the same closed region. It is characterized in that a pixel is selected from each of the detected candidate pixels.

Further, the encoding method according to the present invention is characterized in that candidate pixels close to the direction in which the pixels on the contour line are encoded are selected from the detected candidate pixels.

Further, the coding method according to the present invention calculates the importance coefficient of each candidate pixel detected according to the type and arrangement of the regions on both sides of the candidate pixel in the direction of coding the pixel on the contour line. To do. The candidate pixel having the largest importance coefficient is selected from the calculated importance coefficients of the candidate pixels.

Also, the encoding method according to the present invention detects and accumulates the angle change coefficient indicating the positional relationship between the previous target pixel, the current target pixel, and the next target pixel. The accumulated result is used to determine the type of the closed region of the contour line for which the coding is completed, and the encoded data of the contour line for which the coding is completed is output.

Further, the encoding method according to the present invention is characterized in that, when the next target pixel is selected from the candidate pixels, a contour line image in which the current pixel of interest is deleted from the image is generated. .

Further, the encoding method according to the present invention detects a pixel on the contour line by unidirectionally searching from an arbitrary pixel located at the edge of the image. The first detected pixel on the contour line is set as the start point.

Further, in the coding method according to the present invention, when the pixels on the contour line are not detected, the coding of the contour lines of a plurality of indefinite shape closed regions existing in the image is terminated. Is characterized by.

In order to solve the above-mentioned problems, the coding apparatus according to the present invention extracts the contour lines of a plurality of closed regions of indeterminate shape existing in the image and outputs the coded contour line data. The encoding device comprises: an extracting unit that extracts the contour lines of the plurality of indefinite-shaped closed regions; and an encoding unit that encodes each contour line extracted by the extracting unit. Further, the extraction means is characterized in that the contour line of each closed region is extracted by using the pixel value distribution inside the closed region and the contour line shape of the closed region as the feature amount of each closed region.

Further, in the encoding apparatus according to the present invention, the extraction means binarizes pixels inside the closed area and pixels outside the closed area and closes boundary pixels where the value of the binarized pixel changes. A feature is that a contour line is extracted as a contour pixel of the area.

Further, the coding apparatus according to the present invention is characterized in that the coding means codes each contour line of each closed region.

Further, in the encoding device according to the present invention, the encoding means encodes an arbitrary point on the contour line as a start point, and sequentially traces pixels on the contour line from pixels existing at the start point. The encoding of the contour line is terminated at the time of returning to the start point.

Further, in the coding apparatus according to the present invention, the coding means may select each pixel on the contour line other than the target pixel existing in the window area centered on the target pixel on the contour line as a candidate pixel. And a candidate pixel detecting means for detecting as. Then, an arbitrary candidate pixel among the candidate pixels detected by the candidate pixel detecting means is selected as a target pixel to be encoded next.

Further, in the encoding device according to the present invention, the encoding means is arranged so that pixels on both sides of the target pixel always belong to the same closed region in the direction in which the pixels on the contour line are encoded. The target pixel to be encoded into is selected from each candidate pixel obtained by the candidate detecting means.

Further, in the coding apparatus according to the present invention, the coding means selects candidate pixels close to the direction of coding the pixels on the contour line from the candidate pixels obtained by the candidate detecting means. Is characterized by.

Further, in the encoding device according to the present invention, the encoding means is configured to detect the candidate pixel according to the type and arrangement of regions on both sides of the candidate pixel in the direction of encoding the pixel on the contour line. An importance degree calculating means for calculating the importance degree coefficient of each detected candidate pixel is provided. And
It is characterized in that the candidate pixel having the largest importance coefficient is selected from the importance coefficients of the candidate pixels obtained by the importance calculating means.

Further, in the encoding apparatus according to the present invention, the encoding means detects and accumulates an angle change coefficient indicating a positional relationship between the previous target pixel, the current target pixel, and the next target pixel, and accumulates the angle. A change coefficient accumulating means is provided. Then, the type of the closed region of the contour line for which encoding has been completed is determined based on the accumulation result obtained by the angle change coefficient accumulating means, and the encoded data of the contour line for which encoding has been completed is output. To do.

Further, in the coding apparatus according to the present invention, the coding means generates a contour line image in which the current pixel of interest is deleted from the image when the next target pixel is selected from the candidate pixels. It is characterized in that it is provided with a contour pixel deleting means.

Further, in the encoding device according to the present invention, the encoding means includes a detecting means for detecting a pixel on the contour line by unidirectionally searching from an arbitrary pixel located at an edge of the image. Then, the pixel on the contour line first detected by the detecting means is set as the start point.

Further, in the encoding apparatus according to the present invention, the encoding means includes a plurality of indeterminate shape closed regions existing in the image when the detecting means does not detect a pixel on the contour line. It is characterized in that the encoding of the contour line is finished.

In order to solve the above-mentioned problems, the decoding method according to the present invention uses the pixel value distribution inside a plurality of indefinite-shaped closed regions existing in an image and the contour line shape of the closed region for each closed region. A contour line of each closed region is extracted by using it as a feature amount, and a decoding method for decoding the contour line encoded data sequentially encoded from the outer side contour line is first described. From the contour coded data, all the pixels existing between the contour line and any one pixel inside the closed area are detected in the closed area. Detected as a pixel.
The image is reconstructed based on the detected feature amount and pixels, and the contour line encoded data is decoded for each closed region in the order of encoding.

In order to solve the above-mentioned problems, the decoding apparatus according to the present invention determines the pixel value distribution inside a plurality of indeterminate shaped closed regions existing in an image and the contour shape of the closed region for each closed region. A decoding device for extracting the contour line of each closed region by using it as a feature amount, and decoding the contour coded data sequentially coded from the outer contour line, wherein: A feature amount detecting means for detecting a feature amount, and a pixel existing inside the closed region from the contour line encoded data, and all pixels existing between any one pixel inside the closed region and the contour line. The image forming apparatus includes a detecting unit that detects a pixel in the closed region, and a reconstructing unit that reconstructs the image based on the feature amount detected by the feature amount detecting unit and the pixel detected by the detecting unit. The outline coded data is decoded for each closed region in the order of coding.

[0050]

Embodiments of the present invention will be described below in detail with reference to the drawings.

The encoding method according to the present invention is, for example, as shown in FIG.
The coding apparatus 100 uses 3 CC (3 Component Coding) as shown in FIG.

First, 3CC is a coding method for preserving the contour line portion in the original image, and a coding method for dividing the original image into edge portions, luminance information, and texture information for coding. Is. Further, 3CC is a typical one of the ultra-low bit rate image coding schemes which emphasizes preservation of the contour line of an object.

This encoding apparatus 100 is an application of the encoding apparatus according to the present invention, and as shown in FIG. 1, a luminance information code to which an input image signal DD0 which is a signal of an original image is supplied. Of the edge image reconstruction processing circuit DP3 to which the output of the contour encoding circuit DP2 is supplied, and the output of the edge image reconstruction processing circuit DP3 and the input image signal DD0 are supplied. The differential circuit DP4 and the entropy coding circuit DP5 to which the output of the differential circuit DP4 is supplied.

The output of the contour coding circuit DP2 is
It is also adapted to be supplied to the edge image reconstruction processing circuit DP3. Then, the luminance coded data ED1 output from the luminance information coding circuit DP1 and the contour line coding circuit D
The contour coded data ED2, which is the output of P2, and the texture coded data ED3, which is the output of the entropy coding circuit DP5, are respectively output to the 3CC decoding device described later.

Further, the coding apparatus 100 is provided with an edge area extraction circuit described later.

First, for example, an image including a texture closed region and an edge closed region as shown in FIG. 17 is input to the encoding device 100 as an input image signal DD0.

The luminance information encoding circuit DP1 obtains and encodes a local luminance component indicating global luminance information of the image from the input image signal DD0 input to the encoding device 100, and encodes the encoded local luminance component DD1. It is supplied to the edge image reconstruction processing circuit DP3 and is output as the luminance encoded data ED1.

On the other hand, the contour encoding circuit DP2 extracts the contour portion of the image from the input image signal DD0 by using the edge area extracting circuit and encodes the information of the extracted contour portion. Then, the contour line coding circuit DP2 has information on the coded contour line portion (hereinafter, referred to as contour line information).
DD2 is supplied to the edge image reconstruction processing circuit DP3 and is output as contour line encoded data ED2.

The edge image reconstruction processing circuit DP3 reconstructs an edge image from the local luminance component DD1 from the luminance information encoding circuit DP1 and the contour line information DD2 from the contour line encoding circuit DP2. That is, the edge image reconstruction processing circuit DP3 uses the local luminance component DD1 and the contour line information DD.
A signal of an edge reconstructed image reconstructed from the edge region and pixel values on both sides of the edge region (hereinafter referred to as an edge reconstructed image signal) DD3 is generated based on 2, and the generated edge reconstruction is performed. The image signal DD3 is supplied to the difference circuit DP4.

The difference circuit DP4 takes the difference between the edge reconstructed image signal DD3 from the edge image reconstruction processing circuit DP3 and the input image signal DD0, and the difference signal is used as the texture image signal DD4 in the entropy coding circuit DP4.
5

The entropy coding circuit DP5 codes the texture image signal DD4 from the difference circuit DP4,
The encoded texture image signal DD4 is output as texture encoded data ED3.

A detailed description of the entropy coding circuit DP5 will be given later.

Luminance coded data ED1, contour line coded data ED2 obtained by the coding apparatus 100 as described above,
The texture encoded data ED3 is decoded by the 3CC decoding device 200 as shown in FIG.

The decoding device 200 is a device for carrying out the decoding method according to the present invention, to which the decoding device according to the present invention is applied.

That is, the decoding apparatus 200 is the same as that shown in FIG.
As shown in, the luminance information decoding circuit EP1 to which the luminance encoded data ED1 obtained by the encoding device 100 is supplied
And the output of the luminance information decoding circuit EP1 and the encoding device 1
A contour decoding circuit EP2 to which the contour coded data ED2 obtained in 00 is supplied respectively, and a texture decoding circuit EP3 to which the output of the texture coded data ED3 obtained in the coding device 100 is supplied, Contour decoding circuit E
And an adder circuit EP4 to which the output of P2 and the output of the texture decoding circuit EP3 are respectively supplied. And
The output of the addition circuit EP4 is output as the reconstructed image signal ED7.

First, the luminance information decoding circuit EP1 obtains a local luminance component ED4 by decoding the supplied luminance coded data ED1. Then, the luminance information decoding circuit EP1 supplies the local luminance component ED4 to the contour line decoding circuit EP2.

The contour line decoding circuit EP2 obtains contour line information by decoding the supplied contour line coded data ED2. The contour decoding circuit EP2 reconstructs the edge reconstructed image signal ED5 from the contour information and the local luminance component ED4 from the luminance information decoding circuit EP1, and reconstructs the edge reconstructed image signal ED5. Is supplied to the adder circuit EP4.

On the other hand, the texture decoding circuit EP3 obtains the texture image signal ED6 by decoding the supplied texture encoded data ED3. And
The texture decoding circuit EP3 supplies the texture image signal ED6 to the adding circuit EP4.

A detailed description of the texture decoding circuit EP3 will be given later.

The adder circuit EP4 is a contour line decoding circuit EP.
The edge reconstructed image signal ED5 from 2 and the texture image signal ED6 from the texture decoding circuit EP3 are added, and the addition signal is output as a reconstructed image signal ED7.

As described above, the 3CC decoding device 200
Then, by adding the texture image signal ED6 to the edge reconstructed image signal ED5, it is possible to reproduce a fine change in the original image.

That is, although the edge reconstructed image signal ED5 reconstructed from the edge region and the pixel values on both sides of the edge region includes global features such as the contours and boundaries of the object in the original image, Since almost no information such as a small change in gradation value is included, the decoding device 200 uses the texture image signal ED6, which is a signal of the difference image between the original image and the edge reconstructed image, as a fine pattern such as a pattern of an object. Information including changes and the like is added to the edge reconstructed image signal ED5. This makes it possible to reproduce subtle irregularities such as lawns and forests.

Here, the texture image signal ED6 is the difference circuit DP4 in the encoding device 100 shown in FIG.
Although it corresponds to the texture image signal DD4 obtained in step 1 and is encoded by the entropy encoding circuit DP5, when encoding is performed for each macroblock as in the conventional case, the encoding efficiency becomes poor.

Therefore, the entropy coding circuit DP5 included in the coding apparatus 100 extracts the contour line of the texture closed region and codes only the pixels within the contour line.

The entropy coding circuit DP5 will be specifically described below.

As shown in FIG. 3, the entropy coding circuit DP5 includes an edge area buffer FP0, an edge area expansion circuit FP1 to which the output of the edge area buffer FP0 is supplied, an output of the edge area expansion circuit FP1 and the above-mentioned diagram. 1. The texture area separation circuit FP2 to which the output of the difference circuit DP4 shown in 1 is respectively supplied, the orthogonal transformation circuit FP4 and the contour line listing circuit FP3 to which the outputs of the texture area separation circuit FP2 are respectively provided, and the orthogonal transformation circuit FP4. And a variable length coding circuit FP6 to which the output of the quantizing circuit FP5 and the output of the contour line listing circuit FP3 are respectively supplied, The output buffer FP7 is supplied with the output of the variable length coding circuit FP6.

First, the edge area buffer FP0 stores the edge area signal CD9 from the above-described edge area extraction circuit provided in the encoding apparatus 100.

As shown in FIG. 4, this edge area extracting circuit includes, for example, a horizontal edge strength calculating circuit CP1, a vertical edge strength calculating circuit CP2, and a filtering circuit to which a pixel signal CD0 of a still image is supplied. The processing circuit CP3, the multiplier CP4 supplied with the output of the edge strength calculation circuit CP1, the multiplier CP5 supplied with the output of the edge strength calculation circuit CP2, the output of the multiplier CP4 and the output of the multiplier CP5, respectively. The supplied adder CP7,
A mask area setting circuit CP6 to which the output of the filtering processing circuit CP3 is supplied, a threshold value determining circuit CP8 to which the output of the adder CP7 and the output of the mask area setting circuit CP6 are respectively supplied, and an output of the adder CP7 and a threshold value determining circuit. C
The comparator CP9 is supplied with the output of P8, and the output of the comparator CP9 is stored as the edge area signal CD9 in the edge area buffer BP0 shown in FIG.

First, the pixel signal CD of the original image which is a still image
0 is supplied to the edge strength calculation circuit CP1, the edge strength calculation circuit CP2, and the filtering processing circuit CP3, respectively.

As shown in FIG. 5, the edge strength calculation circuit CP1 has a tap coefficient of In a 3 has a Sobel filter S _H of × 3, using the Sobel filter S _H, obtains the edge strength signals CD1 in the horizontal direction of the edge intensity from the pixel signal CD0 supplied. Then, the edge strength calculation circuit CP1
Supplies the obtained edge strength signal CD1 to the multiplier CP4.

As shown in FIG. 6, the edge strength calculation circuit CP2 has a 3 × 3 Sobel filter S _V having a tap coefficient of 1 0 -1 2 0 -2 1 0 -1. Using the Sobel filter S _V , the edge intensity signal CD2 indicating the edge intensity in the vertical direction is obtained from the supplied pixel signal CD0. Then, the edge strength calculation circuit CP2
Supplies the obtained edge strength signal CD2 to the multiplier CP4.

The multiplier CP4 is an edge strength calculation circuit CP.
The horizontal edge strength signal CD1 from 1 is squared to obtain the horizontal edge strength signal power CD4,
The calculated edge strength signal power CD4 is supplied to the adder CP7.

The multiplier CP5 is an edge strength calculation circuit CP.
The edge strength signal power CD5 in the vertical direction is obtained by squaring the edge strength signal CD2 in the vertical direction from 2.
The calculated edge strength signal power CD5 is supplied to the adder CP7.

The adder CP7 adds the horizontal edge strength signal power CD4 from the multiplier CP4 and the vertical edge strength signal power CD5 from the multiplier CP5, and the addition result is the edge strength of the pixel of interest. Signal (hereinafter,
This is called a pixel edge strength signal. ) Comparator CP as CD7
9 and the threshold value determination circuit CP8.

In this edge area extraction circuit, the edge area and the texture area are considered in consideration of the differences in characteristics between the edge area such as the outline of the object and the boundary line between the objects and the edge area within the texture area. As a feature amount for separation, a pixel value, that is, a gradation value local change characteristic and an index value indicating a global change characteristic are used.

The feature quantity indicating the local change characteristic of the gradation value is the edge strength of the pixel of interest, which is an important feature quantity when determining whether or not it is the edge region, and the global change characteristic is As a feature value indicating the, there is a spatial frequency distribution around the pixel of interest. In this edge area extraction circuit, a feature amount indicating a spatial frequency is used as a feature amount indicating a global change characteristic of a pixel value.

Therefore, the filtering processing circuit CP3
Is, for example, a 3 × 3 Sobel filter S _H shown in FIG. 5 or a high pass filter (HPF) (not shown) having the same configuration as the 3 × 3 Sobel filter S _V shown in FIG. : High Pass Filter), and supplies the output value of the HPF obtained through the pixel signal CD0 supplied to the HPF to the mask area setting circuit CP6 as the output signal CD3.

The HPF used in the filtering processing circuit CP3 is a low pass filter (LPF).
ter) or band pass filter (BPF: Band Pass F)
ilter). The filter used as the HPF, the LPF, or the BPF is not limited to the 3 × 3 Sobel filters S _H and S _V , and the number of taps and the tap coefficient are different, such as a 5 × 5 filter or a 7 × 7 filter. It may be a filter.

Further, as described above, since the feature amount indicating the spatial frequency is used as the feature amount indicating the global change characteristic of the pixel value, specifically, the specific region of the feature amount indicating the spatial frequency is specified. The mask area setting circuit CP6 is a window area for obtaining a local feature amount required when obtaining the feature amount indicating the spatial frequency, for example, as shown in FIG. Mask area M _A1 ,
M _A2 is set around the target pixel C _P. This mask area M
_A1, M _A2, as shown in FIG. 7, a one-dimensional regions on either side of the target pixel C _P in the normal direction of the edge area E _A pixel of interest C _P has. This makes it possible to obtain a change in gradation that greatly changes with the edge area E _A as a boundary from the relationship between the feature amounts on both sides of the edge area E _A.

Therefore, the mask area setting circuit CP6
Is the output signal CD from the filtering processing circuit CP3.
3 for the mask area M _A1 , as shown in FIG.
M _A2 is set, and the feature amount CD6 obtained from the mask areas M _A1 and M _A2 is calculated. Then, the mask area setting circuit CP6 uses the calculated feature amount CD6 as the threshold value determining circuit CP8.
To supply.

The threshold value determining circuit CP8 receives the pixel edge strength signal CD7 from the adder CP7 and the mask area setting circuit C.
Threshold function F with the parameter CD6 from P6 as a parameter
The threshold value is calculated according to (), and the calculated threshold value is used as the threshold signal CD.
8 is supplied to the comparator CP9.

The threshold used in the threshold decision circuit CP8
The function F () separates the texture area and the edge area.
So that we can take a threshold that is optimal for
It is determined by millation. For example, the threshold function
F () is an edge of the pixel of interest, as shown in FIG.
Intensity E and mask area M_A1, M_A2Average value M₁, M_Twoof
It is a function that shows the relationship, with constants a, b, and c, {M₁-(AE-b)} {M_Two-(AE-b)} = c ... represented by Formula 1. Therefore, the threshold function F ()
Is the mask area M_A1, M _A2Average value M₁, M_TwoBy
The edge strength E that becomes the threshold value T is obtained.

The comparator CP9 determines from the pixel edge strength signal CD7 from the adder CP7 and the threshold signal CD8 from the threshold determination circuit CP8 whether the pixel of interest is a texture area or an edge area. , The edge area signal CD which is an output signal when the pixel of interest is a texture area
9 is output as "0", and the edge area signal CD9 is output as "1" when the pixel of interest is in the edge area.

Therefore, the image obtained by using the edge area signal CD9 output from the comparator CP9 as the pixel value of each pixel, that is, the edge image is the edge area extracted by the edge area extraction circuit.

The edge area extraction circuit obtains edge information not only from the luminance signal but also from the color difference signal, and combines the edge information obtained from the luminance signal with the edge information obtained from the color difference signal. By using, it is possible to extract the contours between objects and the boundaries between objects even in a region where the change in luminance is small.

The edge area expansion circuit FP1 outputs the edge area signal CD9 output from the edge area extraction circuit as described above via the edge area buffer FP0 shown in FIG.
Supplied to

The edge area expansion circuit FP1 is for removing from the texture image information of an area having a very large error from the original image existing near the edge area of the edge reconstructed image. Therefore, the edge area expansion circuit FP
In the edge area signal CD9 from the edge area buffer FP0, 1 obtains an expanded edge area using pixels in a certain range from an arbitrary pixel as an expanded mask area. Therefore, the obtained expanded edge area and the texture area other than the expanded edge area are all closed areas. Then, the edge area expansion circuit FP1 supplies the obtained expansion edge area signal (hereinafter referred to as expansion edge area signal) FD1 to the texture area separation circuit FP2.

The texture area separation circuit FP2 supplies the expanded edge area signal FD1 from the edge area expansion circuit FP1.
And the texture image signal DD4, which is the difference signal between the edge-reconstructed image signal DD3 obtained by the edge-image reconstruction processing circuit DP3 and the input image signal DD0.

The texture area separation circuit FP2 uses the expanded edge area signal FD1 from the edge area expansion circuit FP1 to convert the texture image signal DD4 from the difference circuit DP4 to the expanded edge area MA as shown in FIG.
Only the signal of the texture area TA excluding the area is extracted, and the extracted texture image signal is supplied as the texture area signal FD2 to the contour line listing circuit FP3 and the orthogonal transformation circuit FP4, respectively.

At this time, the texture area separation circuit FP2
Sets the pixel value of the texture area signal FD2 of the expanded edge area to "0" and sets the pixel value of the texture area other than the expanded edge area to the texture image signal D from the difference circuit DP4.
Save as D4.

The contour line listing circuit FP3 extracts and lists the contour lines of the texture closed region or the edge closed region for the texture region signal FD2 from the texture region separation circuit FP2.

FIG. 9 specifically shows the internal structure of the contour line listing circuit FP3. The contour line listing circuit FP3 will be specifically described below with reference to FIG.

The contour line list formation circuit FP3 is supplied with the contour line extraction circuit AP1 to which the texture region signal FD2 is supplied from the texture region separation circuit FP2 shown in FIG. 3 and the list to which the output of the contour line extraction circuit AP1 is supplied. Conversion circuit A
P0 and the all list end determination circuit AP10 and all list storage circuit AP to which the outputs of the list forming circuit AP0 are supplied, respectively.
11 and an output buffer AP12 to which the outputs of all list storage circuits AP11 are supplied. Then, the output of the all list end determination circuit AP10 is the list forming circuit AP0.
And the entire list storage circuit AP11, and the output of the output buffer AP12 is supplied to the quantization circuit FP5 and the variable length coding circuit FP6 shown in FIG. .

The list forming circuit AP0 is supplied with the output of the contour line extracting circuit AP1 and the output of the all list end judging circuit AP10, and the output of the start pixel detecting circuit AP2 and the end of all list. Judgment circuit A
Candidate pixel detection circuit AP3 to which each output of P10 is supplied
And the importance calculation circuit AP4 to which the output of the candidate pixel detection circuit AP3 is supplied, the new attention pixel determination circuit AP5 to which the output of the importance calculation circuit AP4 is supplied, and the output of the new attention pixel determination circuit AP5. Cumulative angle calculation circuit AP6
A target pixel deletion circuit AP7 to which the output of the cumulative angle calculation circuit AP6 is supplied, and a single list end determination circuit AP8 and a single list storage circuit AP9 to which the outputs of the target pixel deletion circuit AP7 are respectively supplied. There is. The output of the single list end determination circuit AP8 is supplied to the single list storage circuit AP9 and the candidate pixel detection circuit AP3, respectively, and the output of the single list storage circuit AP9 is the entire list end determination. It is adapted to be supplied to the circuit AP10 and the entire list storage circuit AP11.

First, the texture region signal FD2 obtained by the texture region separation circuit FP2 is used as an input signal to the contour line listing circuit FP3, and this texture region signal FD2 is supplied to the contour line extraction circuit AP1.

The contour line extraction circuit AP1 extracts the contour lines of the texture closed region and the edge closed region from the texture region image corresponding to the supplied texture region signal FD2.

That is, the contour line extraction circuit AP1 is similar to that shown in FIG.
As shown in 0, in the texture area image, the pixel value in the edge closed area E _A is “0”, the pixel value in the texture closed area T _A is “1”, and the pixel values are “0” to “1”. , Or the contours of the texture closed region and the edge closed region are extracted with the pixels O _P1 and O _P2 having the pixel value of “1” at the portion changing from “1” to “0” as contour pixels. As a result, as shown in FIG. 17, the pixels in the texture area image, which are located in solid lines and dotted lines surrounding the texture closed area TA and the edge closed area MA, are extracted as contour lines.

As described above, the contour line extraction circuit AP1
The contour line extracted by means of the list forming circuit A that forms a contour line for each closed curve as a contour line image signal AD1.
Supplied to P0.

In the list forming circuit AP0, first, the start pixel detecting circuit AP2 starts one on an arbitrary closed curve from the pixels existing in the contour line image corresponding to the contour line image signal AD1 from the contour line extracting circuit AP1. Detect pixels.

That is, the start pixel detection circuit AP2 searches in one direction from one pixel at the edge of the contour line image corresponding to the contour line image signal AD1 and sets the first detected contour pixel as the start pixel of the closed curve list. . For example, with respect to the image shown in FIG.
, The pixel D _S at the upper left corner of the image is searched at the search start point P
_Start and search horizontally in order. As a result, the first detected contour pixel, that is, the pixel on the contour line of the texture closed region is detected as the start pixel D _L1 . And
The start pixel detection circuit AP2 supplies the signal of the start pixel D _L1 detected as described above to the candidate pixel detection circuit AP3 as the start pixel signal AD2.

The candidate pixel detection circuit AP3 uses the start pixel detection circuit A to determine the contour line pixel to be listed next.
Based on the start pixel signal AD2 from P2, the start pixel D _L1
The contour line pixel existing next to is detected as a candidate pixel.
When detecting the candidate pixel, the candidate pixel detection circuit AP3 opens a window region of 3 × 3 pixels around the start pixel D _L1 which is the target pixel, and outline pixels other than the start pixel D _L1 existing in the window region. To detect. Then, the candidate pixel detection circuit AP3
Supplies the detected contour pixel signal to the importance calculation circuit AP4 as a candidate pixel signal AD3.

The importance calculating circuit AP4 calculates an importance coefficient for the candidate pixel signal AD3 from the candidate pixel detecting circuit AP3 according to the position in the traveling direction of the list and the pixel arrangement on both sides. That is, the importance calculation circuit AP4
The importance coefficient is calculated from the arrangement of the candidate pixel list in the traveling direction and the regions to which pixels on both sides of the candidate pixel in the traveling direction belong.

When calculating the importance coefficient, as shown in FIG. 12, the importance calculating circuit AP4 firstly calculates the coefficient distribution according to the arrangement in the traveling direction of the list to determine the previous _target pixel P _pr.
And the position of the candidate pixel P _ca with respect to the _target pixel P _cu ,
Values of "1" to "4" are given. Therefore, the candidate pixel is given a larger value as it is located closer to the traveling direction of the list.

When detecting a candidate pixel of the start pixel which is the first detected contour pixel, the coefficient distribution is given a uniform value because the traveling direction of the list is unknown.

Further, when creating the list, the importance calculation circuit AP4 traces the contour pixels so that the pixels on both sides of the list in the traveling direction of the list always belong to the same edge area or texture area. In addition, in order to determine the area inside the closed curve, which will be described later, the importance coefficient based on the pixel arrangement on both sides of the candidate pixel with respect to the target pixel is determined as the pixel on both sides of the candidate pixel in the edge closed area. Yes, or a pixel within the texture closed region.

For example, as shown in FIG.
Importance when the pixel numbers in the 3 × 3 pixel window area W centered on _cu are pixel numbers 1 to 8 and the four pixel numbers on the top, bottom, left and right of the window area W are pixel numbers 9 to 12 The importance coefficient determined by the calculation circuit AP4 will be described.

Table 1 shows the pixel numbers of adjacent pixels on both sides when the candidate pixels are the pixels of pixel numbers 1-8.

[0118]

[Table 1]

Here, the importance calculation circuit AP4 traces contour pixels so that the texture area is located on the left side and the edge area is located on the right side in the traveling direction of the list.

Therefore, in the importance calculation circuit AP4, the pixel arrangement on both sides of the candidate pixel for the target pixel is [left: right] =
In the case of [texture pixel: edge pixel], the importance coefficient is set to “10”, and when [left: right] = [texture pixel: texture pixel], the importance coefficient is set to “5”, and [left:
Right] = [edge pixel: texture pixel], the importance coefficient is set to “0”.

Then, the importance calculation circuit AP4 sums the importance coefficients determined as described above for each individual candidate pixel, and supplies the total result to the new target pixel determination circuit AP5 as the importance coefficient signal AD4. To do.

In order to determine the next candidate pixel, the new pixel-of-interest determination circuit AP5 uses the importance coefficient signal AD4 from the importance calculation circuit AP4 as the candidate pixel having the largest importance coefficient as the next pixel of interest. The determined target pixel is determined as the new target pixel. As a result, among the candidate pixels whose right and left pixel arrangements are correct, the candidate pixel closest to the traveling direction of the list becomes the next target pixel. Then, the new pixel-of-interest determination circuit AP5 outputs the signal of the determined new pixel-of-interest (hereinafter referred to as the new pixel-of-interest signal) AD5 to the cumulative angle calculation circuit AP.
6

Based on the new pixel-of-interest signal AD5 from the new pixel-of-interest determination circuit AP5, the cumulative angle calculation circuit AP6 shows how much the traveling direction of the list is rotated based on the positional relationship between the pixel of interest and the new pixel-of-interest. This is represented by using the angle coefficient as shown in, and the angle coefficient for each pixel is accumulated from the start pixel.

FIG. 14 shows the angle coefficient of each pixel when the importance coefficient is determined as shown in FIG. As shown in FIG. 14, the distribution of the angle coefficient by the arrangement in the traveling direction is determined by the pixel of interest P.
Coefficients "0" to "3" and "-1" to "-3" are given according to the position of the candidate pixel P _ca with respect to _cu and the new pixel of interest P _pr .

Then, the cumulative angle calculation circuit AP6 is made up of the cumulative result of the angle coefficient (hereinafter referred to as cumulative angle information) obtained by using the angle coefficient shown in FIG. 14 and the position information of the new pixel of interest. The output signal AD6 is generated, and the generated output signal AD6 is supplied to the target pixel deleting circuit AP7.

The pixel-of-interest deletion circuit AP7 has the contour obtained by the contour line extraction circuit AP1 based on the output signal AD6 from the cumulative angle calculation circuit AP6, that is, the signal consisting of the cumulative angle information and the position information of the new pixel of interest. The pixel that becomes the pixel of interest is deleted from the image. As a result, the contour pixels incorporated in the list are sequentially deleted from the contour image. Then, the attention pixel deletion circuit AP7 generates an output signal AD7 including the position information of the new attention pixel and the signal of the contour image after the deletion of the new attention pixel, and the generated output signal AD7 is the single list end determination circuit. It is supplied to AP8 and single list storage circuit AP9, respectively.

The single list end judgment circuit AP8 uses the output signal AD7 from the target pixel deletion circuit AP7, that is, the signal consisting of the position information of the new target pixel and the signal of the contour image after the deletion of the new target pixel. When it is determined that the new pixel of interest is the same as the start pixel of the list, it is considered that the list formation for the closed curve is completed, and the single list formation completion signal AD8 indicating the completion is described above as the single list storage circuit AP9. It is supplied to the candidate pixel detection circuit AP3. When the single list end determination circuit AP8 determines that the new pixel of interest is not the same as the start pixel of the list, the single list termination determination circuit AP8 considers that the list of closed curves has not been completed, and indicates that the single curve has not been completed. The completion signal AD8 is supplied to the single list storage circuit AP9 and the above-described candidate pixel detection circuit AP3.

The single list storage circuit AP9 stores the output signal AD7 from the target pixel deletion circuit AP7 as list information.

Further, the single list storage circuit AP9 outputs the single list completion signal A from the single list end judgment circuit AP8.
When D8 is supplied, the stored list information is used as the single list signal AD9, and the all list end determination circuit AP1
0 and supplied to the entire list storage circuit AP11. At this time, the single list storage circuit AP9 performs the above-described region determination processing inside the closed curve.

That is, the single list storage circuit AP9 is
At the time when the listing of each closed region is completed, if the cumulative angle coefficient of the list information is “0” or more, according to the stored list information, the closed curve is traced counterclockwise at the time of listing, The inside of the list information is regarded as a closed curve of a texture closed region, and when the cumulative angle coefficient is “0” or less, the closed curve is traced clockwise in the list formation. Is regarded as a closed curve of the edge closed region, the list information is treated as a single list signal AD9, and the all list end determination circuit AP10 and the all list storage circuit A
Supply to P11 respectively.

When the all list end judging circuit AP10 judges from the single list signal AD9 from the single list storage circuit AP9 that the contour pixels still exist in the contour image, the complete list formation is not completed. The signal AD10 is supplied to the start pixel detection circuit AP2 and the candidate pixel detection circuit AP3.

As a result, the start pixel detection circuit AP2 is
When the all-listing completion signal AD10 is supplied from the all-list end determination circuit AP10, the start pixel is detected as described above. Further, the candidate pixel detection circuit AP3 is
When the all-listing completion signal AD10 is supplied from the all-list end determination circuit AP10, candidate pixels for determining the contour line pixel to be listed next are detected. Therefore, in this case, listing is started again.

Further, the all list end judgment circuit AP10 is
Single list signal AD9 from single list storage circuit AP9
Thus, when it is determined that there are no contour pixels in the contour image, the all-listing completion signal AD10 is supplied to the all-list storage circuit AP11.

The all-list storage circuit AP11 stores the single-list signal AD9 from the single-list storage circuit AP9 as all-listing information. And the entire list storage circuit AP
When the all list completion signal AD10 is supplied from the all list end determination circuit AP10, the output buffer AP outputs the stored all list information as an output signal AD11.
Output via 12.

Therefore, as a result of the above-described listing by the contour list listing circuit FP3, for example, in the image shown in FIG. 11, a list of closed curves is generated in the order of List1 to List4. Also, List1,
Since List3 and List4 are listed in the counterclockwise direction, they become a list surrounding the texture closed area.
Is a list enclosing an edge closed region because it is listed clockwise.

The output signal AD11 output from the output buffer AP12 is the signal supplied from the contour line listing circuit FP3 shown in FIG. 3 to the quantization circuit FP5 and the variable length coding circuit FP6, that is, the contour. It becomes the line listing signal FD3.

On the other hand, the orthogonal transformation circuit FP4 uses a wavelet (Wave) using a 5-3 filter for the texture area signal FD2 from the texture area separation circuit FP2.
let) Convert. The orthogonal transform circuit FP4 is configured to be able to obtain an orthogonal transform coefficient in which pixel position information is stored by performing this wavelet transform. Then, the orthogonal transform circuit FP4 supplies the orthogonal transform coefficient signal FD4 obtained by performing the wavelet transform to the quantization circuit FP5.

The quantization circuit FP5 is the orthogonal transformation circuit FP4.
Quantize the orthogonal transform coefficient signal FD4 from At this time,
The quantization circuit FP5 adaptively changes the quantization step size used for quantization in accordance with the characteristics inside each texture closed region based on the contour line listing signal FD3 from the contour list listing circuit FP3. Let For example, when the amplitude value of the orthogonal transform coefficient in the area is large, a large quantization step size is used, and when the amplitude value of the orthogonal transform coefficient in the area is small, a small quantization step size is used.

As described above, the quantization circuit FP5 is
The quantization step size is adjusted according to the feature amount extracted from the pixel information in each list, and quantization is performed using the adjusted quantization step size. Then, the quantization circuit FP5
Is a variable length coding circuit F as a quantized orthogonal transformation coefficient signal FD5 obtained by quantizing the orthogonal transformation coefficient signal FD4.
Supply to P6.

The variable length coding circuit FP6 codes and codes the quantized orthogonal transform coefficient signal FD5 from the quantization circuit FP5 based on the contour line listing signal FD3 from the contour line listing circuit FP3. The obtained code is output as texture encoded data FD6 via the output buffer FP7.

Therefore, the texture coded data FD6 output from the output buffer FP5 becomes the signal DD5 output from the entropy coding circuit DP5 shown in FIG. 1, and the texture coded data ED3 shown in FIG. 2 is obtained. It will be decrypted by the decryption device 200.

As described above, in the encoding device 100, the entropy encoding circuit DP5 extracts the outlines of the texture closed region and the edge closed region in the texture image and lists them, so that the information in the texture closed region is stored. Only can be encoded. Therefore, the coding efficiency can be improved. Further, when a moving image is used as a processing target image, the texture closed region in each frame is associated with the texture closed regions in the preceding and subsequent frames, and can be used for motion compensation processing of the texture closed region. Thereby, the coding efficiency can be further improved.

The above-mentioned FIG. 2 for decoding the texture coded data ED3 obtained by the above-described coding device 100 is used.
Texture decoding circuit EP of the decoding device 200 shown in FIG.
3 will be specifically described.

The texture decoding circuit EP3 shown in FIG.
5, the entropy encoding circuit DP5 of the encoding apparatus 100 is provided.
Decode the texture encoded data ED3 obtained in
The contour line list signal of the decoded texture coded data ED3, that is, the contour line list signal FD obtained by the contour line list circuit FP3 as shown in FIG.
3 (= AD11) is supplied to the list decoding processing circuit.

The list decoding processing circuit is the same as that shown in FIG.
As shown in, the list start header detection circuit GP1 to which the contour line listing signal FD3 is supplied through the contour list frame buffer GP9, and the information read circuit GP0 to which the output of the list start header detection circuit GP1 is supplied, The contour reconstructing circuit G to which the output of the information reading circuit GP0 is supplied
P5, output of contour line reconstructing circuit GP5 and output of information reading circuit GP0 are supplied respectively, closed region internal reconstructing circuit GP6, and output of closed region internal reconstructing circuit GP6 are supplied. The circuit GP7, the reconstructed image storage buffer GP8 to which the output of the all-list end header detection circuit GP7 and the output of the closed region internal reconstruction circuit GP6 are respectively supplied, and the output buffer to which the output of the reconstructed image storage buffer GP8 is supplied. And GP10. The output of the all list end header detection circuit GP7 is also supplied to the list start header detection circuit GP1.

The information read circuit GP0 includes a list information read circuit GP2 to which the output of the list start header detection circuit GP1 is supplied, a pixel position information read circuit GP3 to which the output of the list information read circuit GP2 is supplied, and a pixel. A list end header detection circuit GP4 to which the output of the position information reading circuit GP3 is supplied. Then, the output of the list end header detection circuit GP4 and the pixel position information read circuit GP
3 is supplied to the contour line reconstructing circuit GP5 and the pixel position information reading circuit GP3, and the output of the list information reading circuit GP2 is the closed region internal reconstructing circuit GP6.
To be supplied.

First, the list start header detection circuit GP1
Detects a start header G1 indicating the start of each list from the contour line listing signal FD3 from the contour list frame buffer GP9, and detects the detected start header G1 and the supplied contour listing signal FD3. GP
Feed to 2.

The list information reading circuit GP2 is based on the start header G1 from the list start header detecting circuit GP1.
Information such as the number of pixels in the list and the internal area type existing after the start header is read from the contour line listing signal FD3 from the list start header detection circuit GP1, and the read information is used as the list internal information GD2 to reconstruct the closed area inside. The list start header detection circuit G is supplied to the circuit GP6.
The contour line listing signal FD3 from P1 is supplied to the pixel position information reading circuit GP3.

The pixel position information reading circuit GP3 reads the position information of the pixels of each list from the contour line listing signal FD3 from the list information reading circuit GP2, and reads the read position information as list pixel position information GD3. The list information read circuit G is supplied to the configuration circuit GP5.
The contour line listing signal FD3 from P2 is supplied to the list end header detection circuit GP4.

The end-of-list header detection circuit GP4 receives the contour line listing signal FD from the pixel position information reading circuit GP3.
3, that is, until the list end header indicating the end of the list is detected from the contour line listing signal FD3 supplied to this list decoding processing circuit, the list end header signal GD
4 is supplied to the pixel position information reading circuit GP3 and the contour reconstructing circuit GP5 with "false = 0", and when the list end header is detected, the list end header signal GD4 is set with "true = 1". The reading circuit GP3 and the contour reconstructing circuit GP5 are supplied respectively. Further, the list end header detection circuit GP4 supplies the contour line listing signal FD3 from the pixel position information reading circuit GP3 to the contour line reconstruction circuit GP5.

As a result, the pixel position information reading circuit GP3
Repeats the above-described read processing of the list pixel position information GD3 until the list end header signal GD4 from the list end header detection circuit GP4 becomes “true = 1”.

When the list end header signal GD4 from the list end header detection circuit GP4 is "true = 1", the contour line reconstruction circuit GP5 outputs the list pixel position information GD3 from the pixel position information read circuit GP3. And the contour line listing signal FD3 from the pixel position information reading circuit GP3, the contour line is reconstructed, the contour line shape signal GD5 obtained by the reconstruction, and the contour line list from the pixel position information reading circuit GP3. The converted signal FD3 to the closed region internal reconstruction circuit GP6
To supply.

The closed area internal reconstructing circuit GP6 has list internal information GD2 which is information such as the number of pixels in the list from the list information reading circuit GP2 and the kind of internal area, and the contour line shape signal GD5 from the contour reconstructing circuit GP5. , And the contour line listing signal FD3 from the contour reconstructing circuit GP5,
If it is determined that the inside is the texture area, fill the inside pixels with the pixel values restored from the texture coefficient,
When it is determined that the inside is the edge area, the image of the texture closed area is reconstructed by filling the inside pixels with the value of “0”.

Here, in the closed area internal reconstruction circuit GP6, when detecting a pixel inside the closed area, an arbitrary one pixel inside the closed area is set in each contour line, that is, in the case of a texture area, Pixels existing on the left side with respect to the traveling direction of the list are set, and in the case of an edge area, pixels existing on the right side with respect to the traveling direction of the list are set, and exist between the target pixel and the contour pixel. All pixels to be filled are filled as a closed area.

At this time, for example, the Li shown in FIG.
Even if different closed curves exist inside the closed curve like st1 and List2, the inside of the texture closed curve is filled with the texture signal and the inside of the edge closed curve is set to the pixel value in order to always decode from the outside in the order of the list numbers. "0"
The process of filling in is repeated. This allows
It is possible to completely restore even inside the overlapping closed regions.

Then, the closed region internal reconstruction circuit GP6 is
The image of the texture closed region obtained by the reconstruction as described above is stored as the reconstructed image GD6 in the reconstructed image storage buffer GP8, and at the same time, the contour line reconstruction circuit GP5.
The contour list forming signal FD3 from the above is supplied to the all list end header detecting circuit GP7.

The all-list end header detection circuit GP7 receives the contour line listing signal F from the closed region internal reconstruction circuit GP6.
If it is determined by D3 that the decoding of the entire list is not completed, the entire list end header signal GD7 is set to "false =
0 "to the list start header detection circuit GP1 and the reconstructed image storage buffer GP8, and when it is determined that the decoding of the entire list is completed, the entire list end header signal GD7 is set to" true = 1 ". It is supplied to the start header detection circuit GP1 and the reconstructed image storage buffer GP8.

As a result, the list start header detection circuit G
P1 detects the list start header as described above until the all-list end header signal GD7 from the all-list end header detection circuit GP7 becomes “true = 1”. That is, the list start header detection circuit GP1 receives the all list end header signal G from the all list end header detection circuit GP7.
When D7 is "false = 0", the list start header for decoding the next list is detected. And
The closed region is reconstructed again.

Also, the reconstructed image storage buffer GP8 is
When the all-list end header signal GD7 from the all-list end header detection circuit GP7 is “true = 1”, the reconstructed images accumulated until the all-list end header signal GD7 becomes “true = 1”. The GD6 is output as a reconstructed image GD8 of the texture closed region via the output buffer GP10.

The reconstructed image GD8 output from the output buffer GP10 is the texture image signal E output from the texture decoding circuit EP3 shown in FIG.
Then, the texture image signal ED6 is supplied to the addition circuit EP4.

As described above, the texture-encoded data ED3 obtained by the encoding device 100 is always the outside even if different closed curves exist inside the closed curves like List1 and List2 shown in FIG. Since the data are listed from the closed curve, when decoding the texture encoded data ED3 by the decoding device 200, the inside of the texture closed curve is filled with the texture signal and the inside of the edge closed curve is By filling the pixel value “0”, the inside of the overlapping closed region can be completely restored.

[0162]

According to the encoding method of the present invention, the contour lines of a plurality of indeterminately shaped closed regions existing in an image are characterized by the pixel value distribution inside the closed regions and the contour lines of the closed regions as the characteristics of each closed region. Use as quantity and extract. Then, the contour line encoded data obtained by encoding each extracted contour line is output. As a result, only the pixels inside the closed region can be encoded. Therefore, the coding efficiency can be improved. When the above image is a moving image, the closed region in each frame is associated with the closed regions in the preceding and succeeding frames, and can be used for motion compensation processing of the closed region. Thereby, the coding efficiency can be further improved.

In the encoding method according to the present invention, the pixels inside the closed area and the pixels outside the closed area are binarized. And 2
A contour line is extracted by using a boundary pixel where the value of the binarized pixel changes as a contour pixel of the closed region. As a result, it is possible to extract the contour lines of the plurality of irregularly shaped closed regions.

In the coding method according to the present invention, coding is performed for each contour line of each closed area. As a result, it is possible to obtain contour line encoded data in which the contour lines of the plurality of closed regions of indeterminate shape are encoded for each contour line of each closed region.

Further, in the encoding method according to the present invention, an arbitrary point on the contour line is set as a start point, and the pixels on the contour line are sequentially traced from the pixel existing at the start point for encoding. Then, the encoding of the contour line is terminated when the point returns to the start point. As a result, the contour line of each closed region can be accurately encoded.

Further, in the encoding method according to the present invention, pixels on the contour line other than the target pixel existing in the window area centered on the target pixel on the contour line are detected as candidate pixels. Then, an arbitrary candidate pixel among the detected candidate pixels is selected as a target pixel to be encoded next. As a result, the pixel to be coded next can be coded.

Further, in the encoding method according to the present invention, the target pixel to be encoded next such that the pixels on both sides of the target pixel in the encoding direction of the pixel on the contour line always belong to the same closed region. Is selected from each candidate pixel. With this, the pixels on both sides of the target pixel in the encoding direction can always belong to the same closed region.

Further, in the coding method according to the present invention, candidate pixels close to the direction of coding the pixels on the contour line are selected from each candidate pixel. This makes it possible to code an appropriate pixel to be coded next.

In the coding method according to the present invention, the importance coefficient of each candidate pixel detected is calculated according to the type and arrangement of the regions on both sides of the candidate pixel in the direction in which the pixels on the contour line are coded. To do. Then, the candidate pixel having the largest importance coefficient is selected from the calculated importance coefficients of the candidate pixels. As a result, the most appropriate pixel to be coded next can be coded.

Further, in the encoding method according to the present invention, the angle change coefficient indicating the positional relationship between the previous target pixel, the current target pixel and the next target pixel is detected and accumulated. Then, the type of the closed region of the contour line for which coding has been completed is discriminated from the accumulated result, and the coded data of the contour line for which the coding has been completed is output. Thereby, the encoded data of the contour line can be obtained for each closed region.

Further, in the encoding method according to the present invention, when the next target pixel is selected from the candidate pixels, a contour line image in which the current pixel of interest is deleted from the image is generated.
This makes it possible to obtain a contour line image in which coded contour pixels are sequentially deleted from the image.

Further, in the coding method according to the present invention, the pixels on the contour line are detected by unidirectionally searching from arbitrary pixels located at the edge of the image. The first detected pixel on the contour line is set as the start point. This allows
It is possible to sequentially code from the contour line of the outer closed region.

Further, in the coding method according to the present invention, when the pixels on the contour line are not detected, the coding of the contour lines of a plurality of indeterminately shaped closed regions existing in the image is completed. With this, it is possible to accurately encode the contour lines of a plurality of closed regions having an indefinite shape for each contour line of each closed region.

In the encoding device according to the present invention, the extraction means uses the pixel value distribution inside the closed region and the contour line shape of the closed region as the feature amount of each closed region, so that a plurality of indefinite points existing in the image are detected. The contour line of the closed region of the shape is extracted. Then, the encoding means outputs the contour line encoded data obtained by encoding each contour line extracted by the extraction means. As a result, only the pixels inside the closed region can be encoded. Therefore, the coding efficiency can be improved. When the above image is a moving image,
By associating the closed area in each frame with the closed areas in the preceding and following frames, the closed area can be used for the motion compensation processing of the closed area. Thereby, the coding efficiency can be further improved.

Further, in the encoding apparatus according to the present invention, the extraction means binarizes the pixels inside the closed area and the pixels outside the closed area and closes the boundary pixels where the value of the binarized pixel changes. A contour line is extracted as a contour pixel of the area. As a result, it is possible to extract the contour lines of the plurality of irregularly shaped closed regions.

In the coding apparatus according to the present invention, the coding means codes each contour line of each closed region. As a result, it is possible to obtain contour line encoded data in which the contour lines of the plurality of closed regions of indeterminate shape are encoded for each contour line of each closed region.

Further, in the coding apparatus according to the present invention, the coding means sets an arbitrary point on the contour line as a start point.
The pixels on the contour line are sequentially traced from the pixel existing at the starting point to be coded, and the coding of the contour line is finished when the pixel returns to the starting point. As a result, the contour line of each closed region can be accurately encoded.

Further, in the encoding device according to the present invention, the candidate pixel detecting means of the encoding means is arranged on the contour line other than the target pixel existing in the window area centered on the target pixel on the contour line. Each pixel of is detected as a candidate pixel.
Then, an arbitrary candidate pixel among the candidate pixels detected by the candidate pixel detecting means is selected as a target pixel to be encoded next. As a result, the pixel to be coded next can be coded.

Further, in the coding apparatus according to the present invention, the coding means is arranged so that pixels on both sides of the target pixel always belong to the same closed area in the direction of coding the pixels on the contour line. The target pixel to be encoded into is selected from each candidate pixel obtained by the candidate detecting means. This allows
Pixels on both sides of the target pixel with respect to the encoding direction can always belong to the same closed region.

In the coding apparatus according to the present invention, the coding means selects candidate pixels close to the direction in which the pixels on the contour line are coded from among the candidate pixels obtained by the candidate detecting means. This makes it possible to code an appropriate pixel to be coded next.

Further, in the encoding apparatus according to the present invention, the importance degree calculating means of the encoding means is adapted to the type and arrangement of the regions on both sides of the candidate pixel with respect to the encoding direction of the pixel on the contour line. The importance coefficient of each candidate pixel detected by the candidate pixel detecting means is calculated. Then, the candidate pixel having the largest importance coefficient is selected from the importance coefficients of the candidate pixels obtained by the importance calculating means. As a result, the most appropriate pixel to be coded next can be coded.

Further, in the encoding apparatus according to the present invention, the angle change coefficient accumulating means of the encoding means is the previous target pixel,
The angle change coefficient indicating the positional relationship between the current target pixel and the next target pixel is detected and accumulated. Then, the type of the closed region of the contour line for which the encoding has been completed is determined based on the accumulation result obtained by the angle change coefficient accumulating means, and the encoded data of the contour line for which the encoding has been completed is output. Thereby, the encoded data of the contour line can be obtained for each closed region.

In the coding apparatus according to the present invention, the contour pixel deleting means of the coding means deletes the current pixel of interest from the image when the next target pixel is selected from the candidate pixels. Generate a contour image. This makes it possible to obtain a contour line image in which coded contour pixels are sequentially deleted from the image.

Further, in the encoding apparatus according to the present invention, the detecting means of the encoding means detects the pixels on the contour line by unidirectionally searching from arbitrary pixels located at the edge of the image. Then, the pixel on the contour line that is first detected by the detection means is set as the start point. As a result, it is possible to sequentially perform coding from the contour line of the outer closed region.

Further, in the encoding apparatus according to the present invention, the encoding means may include a plurality of irregularly shaped closed regions existing in the image when the detecting means does not detect a pixel on the contour line. The encoding of the contour line of is finished. With this, it is possible to accurately encode the contour lines of a plurality of closed regions having an indefinite shape for each contour line of each closed region.

In the decoding method according to the present invention, first, by using the pixel value distributions inside a plurality of indefinite-shaped closed regions existing in an image and the contour shape of the closed regions as the feature amount of each closed region, The contour line of the closed region is extracted, and the feature amount is detected from the contour line coded data sequentially coded from the outer contour line. Next, from the contour line encoded data, a pixel existing inside the closed region is detected as all pixels existing between any one pixel inside the closed region and the contour line as pixels within the closed region. . Then, the image is reconstructed based on the detected feature amount and pixels, and the contour coded data is decoded for each closed region in the order of coding. As a result, the pixels in the overlapping closed region can be completely restored.

In the decoding device according to the present invention, the feature amount detecting means uses the pixel value distribution inside a plurality of indeterminate shaped closed regions existing in the image and the contour shape of the closed regions as the feature amount of each closed region. The outline of each closed region is extracted by using
The feature amount is detected from the contour line encoded data sequentially encoded from the outer contour line. The detecting means detects pixels existing in the closed region from the contour line encoded data, and all pixels existing between any one pixel in the closed region and the contour line as pixels in the closed region. To do. Then, the reconstructing unit reconstructs the image based on the feature amount detected by the feature amount detecting unit and the pixel detected by the detecting unit, and the contour for each closed region in the order of encoding. Decode the line encoded data. This allows
Pixels in the overlapping closed areas can also be completely restored.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the configuration of a 3CC coding apparatus to which a coding method according to the present invention is applied.

FIG. 2 is a block diagram showing a configuration of a 3CC decoding device to which a decoding method according to the present invention is applied.

FIG. 3 is a block diagram showing a configuration of an entropy encoding circuit of the encoding device.

FIG. 4 is a block diagram showing a configuration of an edge area extraction circuit of the encoding device.

FIG. 5 is a diagram for explaining a Sobel filter used for extracting a horizontal edge in the filtering processing circuit of the edge area extraction circuit.

FIG. 6 is a diagram for explaining a Sobel filter used for extracting a vertical edge in the filtering processing circuit.

FIG. 7 is a diagram for explaining a one-dimensional mask area used for obtaining a global characteristic of a gradation value in the mask area setting circuit of the edge area extraction circuit.

FIG. 8 is a diagram for explaining a threshold decision function used in the threshold decision circuit of the edge area extraction circuit.

FIG. 9 is a block diagram showing a configuration of a contour line listing circuit of the entropy coding circuit.

FIG. 10 is a diagram for explaining the extraction processing of the edge closed region and the texture closed region in the contour line extraction circuit of the contour line listing circuit.

FIG. 11 is a diagram for explaining a start point determined by a start pixel detection circuit of the contour line listing circuit.

FIG. 12 is a diagram for explaining an importance coefficient calculated by an importance calculation circuit of the contour line listing circuit.

FIG. 13 is a diagram for explaining pixel numbers of neighboring pixels when a pixel of interest is the center when calculating the importance coefficient by the importance calculating circuit.

FIG. 14 is a diagram for explaining an angle coefficient calculated by a cumulative angle calculation circuit of the contour line listing circuit.

FIG. 15 is a block diagram showing a configuration of a texture decoding circuit of the decoding device.

FIG. 16 is a block diagram showing a configuration of a conventional texture image encoding circuit.

FIG. 17 is a diagram for explaining a texture closed region and an edge closed region.

FIG. 18 is a diagram for explaining a texture closed region block obtained by the encoding circuit.

[Description of Codes] DP3 Edge image reconstruction processing circuit DP4 Difference circuit DP5 Entropy coding circuit FP0 Edge area buffer FP1 Edge area expansion circuit FP2 Texture area separation circuit FP3 Contour line listing circuit FP4 Orthogonal transformation circuit FP5 Quantization circuit FP6 Variable Long coding circuit FP7 output buffer

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI Technical display location H04N 5/208 H04N 1/40 B 7/30 103A 9/68 103 7/133 Z

Claims (26)

[Claims] 1. An encoding method for extracting contour lines of a plurality of indeterminate shape closed regions existing in an image and outputting the encoded contour line encoded data, wherein the plurality of indeterminate shape closed regions are provided. The encoding method is characterized in that the contour line is extracted using the pixel value distribution inside the closed region and the contour shape of the closed region as the feature amount of each closed region, and each extracted contour line is encoded. 2. A pixel inside a closed area and a pixel outside the closed area are binarized, and a contour line is extracted by using a boundary pixel where the value of the binarized pixel changes as a contour pixel of the closed area. The encoding method according to claim 1, wherein 3. The encoding method according to claim 1, wherein encoding is performed for each contour line of each closed region. 4. An arbitrary point on the contour line as a starting point,
4. The encoding method according to claim 3, wherein the pixels on the contour line are sequentially traced from the pixels existing at the start point to be encoded, and the encoding of the contour line is terminated when the pixel returns to the start point. . 5. A pixel on the contour line other than the target pixel existing in a window area centered on the target pixel on the contour line is detected as a candidate pixel, and any candidate pixel among the detected candidate pixels is detected. 5. The encoding method according to claim 4, wherein is selected as a target pixel to be encoded next. 6. A target pixel to be coded next is selected from each candidate pixel so that pixels on both sides of the target pixel in the direction of coding the pixel on the contour line always belong to the same closed region. The encoding method according to claim 5, wherein: 7. The encoding method according to claim 5, wherein a candidate pixel close to the direction of encoding the pixel on the contour line is selected from the detected candidate pixels. 8. A candidate obtained by calculating the importance coefficient of each candidate pixel detected according to the type and arrangement of regions on both sides of the candidate pixel with respect to the encoding direction of the pixel on the contour line. The encoding method according to claim 5, wherein a candidate pixel having the largest importance coefficient is selected from the importance coefficients of the pixels. 9. An angle change coefficient indicating a positional relationship between a previous target pixel, a current target pixel, and a next target pixel is detected and accumulated, and a closed region of a contour line of which encoding is completed is accumulated according to the accumulated result. The encoding method according to claim 5, wherein the encoded data of the contour line for which the encoding has been completed is determined and the encoded data is output. 10. The encoding method according to claim 5, wherein when the next target pixel is selected from the candidate pixels, a contour line image in which the current pixel of interest is deleted from the image is generated. 11. The pixel on the contour line is detected by unidirectionally searching from any pixel located on the edge of the image, and the first detected pixel on the contour line is used as the starting point. The encoding method according to claim 4. 12. The encoding of contour lines of a plurality of closed regions of indeterminate shape existing in the image is terminated when a pixel on the contour line is not detected. Encoding method. 13. A coding device for extracting contour lines of a plurality of indeterminate shape closed regions existing in an image and outputting the encoded contour line coded data, the plurality of irregular shape closed regions. And an encoding means for encoding each contour line extracted by the extracting means, wherein the extracting means is a pixel value distribution inside the closed region and a contour line shape of the closed region. An encoding device characterized by extracting the contour line of each closed region by using as a feature amount of each closed region. 14. The extraction means binarizes pixels in a closed area and pixels outside the closed area, and extracts a contour line by using a boundary pixel where the value of the binarized pixel changes as a contour pixel of the closed area. The encoding device according to claim 13, wherein 15. The encoding apparatus according to claim 13, wherein the encoding means encodes each contour line of each closed region. 16. The encoding means encodes an arbitrary point on the contour line as a start point by sequentially tracing pixels on the contour line from pixels existing at the start point, and at the time of returning to the start point. The coding apparatus according to claim 15, wherein coding of the contour line is terminated. 17. The encoding means includes candidate pixel detection means for detecting, as candidate pixels, each pixel on the contour line other than the target pixel existing in a window area centered on the target pixel on the contour line. 17. The encoding apparatus according to claim 16, wherein any of the candidate pixels detected by the candidate pixel detecting means is selected as a target pixel to be encoded next. 18. The candidate for the target pixel to be coded next such that pixels on both sides of the target pixel always belong to the same closed region in the direction of coding the pixel on the contour line. 18. The coding apparatus according to claim 17, wherein the coding device is selected from each candidate pixel obtained by the detection means. 19. The encoding unit according to claim 17, wherein the encoding unit selects a candidate pixel close to a direction in which the pixels on the contour line are encoded, from each candidate pixel obtained by the candidate detecting unit. Encoding device. 20. The importance of each candidate pixel detected by the candidate pixel detecting means according to the type and arrangement of regions on both sides of the candidate pixel with respect to a direction in which the pixels on the contour are encoded, 18. A candidate pixel having a largest importance coefficient is selected from the importance coefficients of the candidate pixels obtained by the importance calculation means, the candidate pixel having a highest importance coefficient is selected. Encoding device described. 21. The encoding means includes an angle change coefficient accumulating means for detecting and accumulating an angle change coefficient indicating a positional relationship between a previous target pixel, a current target pixel and a next target pixel, and the angle 18. The coded data of the contour line for which the coding has been completed is output by determining the type of the closed region of the contour line for which the coding has been completed based on the accumulation result obtained by the change coefficient accumulating means. Encoding device described. 22. The encoding means comprises contour pixel deleting means for generating a contour line image in which the current pixel of interest is deleted from the image when the next target pixel is selected from the candidate pixels. 18. The encoding device according to claim 17, which is characterized in that 23. The encoding means comprises a detection means for detecting a pixel on the contour line by unidirectionally searching from an arbitrary pixel located at an edge of the image, and first detected by the detection means. The encoding device according to claim 16, wherein a pixel on the contour line is set as the starting point. 24. The encoding means terminates the encoding of the contour lines of a plurality of irregularly shaped closed regions existing in the image when the detecting means does not detect pixels on the contour line. 24. The encoding device according to claim 23, wherein: 25. A contour line of each closed region is extracted by using a pixel value distribution inside a plurality of indefinite-shaped closed regions existing in an image and a contour line shape of the closed region as a feature amount of each closed region, A decoding method for decoding contour coded data sequentially coded from an outer contour line, wherein the feature amount is detected from the contour coded data, and the closed region is obtained from the contour coded data. Pixels existing inside are detected as all pixels existing between any one pixel inside the closed area and the contour line as pixels within the closed area, and the image is displayed based on the detected feature amount and pixels. A decoding method comprising reconstructing and decoding the contour coded data for each closed region in the order of coding. 26. A contour line of each closed region is extracted by using a pixel value distribution inside a plurality of indeterminate shaped closed regions existing in an image and a contour line shape of the closed region as a feature amount of each closed region, A decoding device for decoding contour coded data sequentially coded from an outer contour line, comprising: feature amount detection means for detecting the feature amount from the contour line coded data; Detection means for detecting pixels existing in the closed region from the data as all pixels existing between the arbitrary one pixel in the closed region and the contour line as pixels in the closed region; A reconstructing means for reconstructing the image based on the feature amount detected by the means and the pixels detected by the detecting means, and decoding the contour line encoded data for each closed region in the order of encoding. To turn Decoding apparatus according to symptoms.