JPH09274662A - Method for encoding and decoding contour - Google Patents

Abstract

PROBLEM TO BE SOLVED: To efficiently encode/decode only a contour graphic by calculating the intersection of a straight line and a contour crossing a base axis at a prescribed angle, encoding the distance between the respective intersections and the base axis and defining the result as a contour encode signal. SOLUTION: A blocker 2 divides the contour graphic into rectangular blocks and segments them. While defining one side of the segmented rectangular blocks as the base axis, a sampler 3 divides the block into (n+1) pieces with (n) pieces of straight lines (corresponding to seven straight lines to which arrows are added) provided along that base axis. Assuming that the block is to be divided by a vertical line with a horizontal line at the lower end of the block as the base axis, for example, (n) pieces of vertical lines inside the block become the contour lines of division. Concerning the respective contour lines, a contour detector 4 calculates the number of true values positioned on the contour lines from the base axis. These values are corresponding to [x0-x7]. An orthogonal transformer 5 simultaneously orthogonally transforms [x0, x1,..., x7], variable length encoding of the orthogonally transformed component is performed by an encoder 6, and the encoded signal is provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a contour coding method and a contour decoding method for efficiently coding and decoding contour figures used for various image processing and image coding, and a contour coding method using the method. The present invention relates to a recording medium in which a decoding device and its method are recorded.

[0002]

2. Description of the Related Art As a conventional coding method for contour figures, there is a famous method called chain coding. In this method, the contour is divided into horizontal, vertical, and diagonal minute vectors, and the symbols corresponding to the minute vectors are encoded. Reversible encoding is possible if the precision of the minute vector is one pixel, and if the precision of the minute vector is rough, the number of bits can be saved more than in the case of lossy but lossless.

[0003]

However, the characteristic of the chain coding is that reversible coding can be realized, and the coding efficiency is not good in the irreversible coding. Therefore, even if a slight distortion is allowed, it is not suitable for image coding for the purpose of significant bit saving. Therefore, contour coding using curve approximation (Japanese Patent Laid-Open Nos. 58-134745 and 58-134745) Etc. are also proposed. This is to encode the contour figure with the position of the characteristic point on the contour and the curve parameter for curve approximation of the contour between the characteristic points, but the amount of calculation for deriving the curve is enormous and the encoding is performed. There is a problem that the number of bits cannot be controlled to a predetermined number of bits. Therefore, ISO / IEC JTC1 / SC29 / WG
As introduced in 11 N1064 etc., a method of approximating a contour figure by a polygon instead of a curve and encoding the difference between the polygon and the actual contour is considered.

However, the method of approximating a contour with a polygon is also characterized in that the vertices of the polygon are on the contour as in the case of curve approximation, and the contour has a complicated shape and many vertex coding is required. In graphics, many bits are required to encode vertices, and the encoding efficiency is poor. As described above, there has been a problem to improve the efficiency of the encoding / decoding method for the contour figure.

In addition, the contour is often treated at the same time as the color signal of the figure having the contour, and in the current mainstream of encoding, JPEG or JPEG, the color signal is divided into blocks and encoded. Therefore, if the contour figure cannot be encoded in the same block unit as the color signal, from the viewpoint of an encoding device that encodes all contours and color images, many devices are required to encode the contour figure. Only needed.

The present invention particularly provides a method for efficiently encoding and decoding only a contour figure, and in addition, a contour figure encoding method applicable to an image encoding method including a current contour figure. It is intended to provide.

[0007]

In order to achieve the above object, the present invention has the following constitution.

A first aspect of the present invention is a contour coding method for coding a contour figure, wherein a straight line selectively arranged around the contour figure is used as a base axis, and n lines intersecting the base axis at a predetermined angle. Of the straight line and the contour, and the distance {x1, x2, ..., xn} between each intersection and the base axis is coded to obtain a contour coded signal. Further, the decoding method is such that the encoded signal is decoded to derive {x1, x2, ..., xn}, and the distance from the base axis on n straight lines intersecting the base axis at a predetermined angle. N points where {x1, x2, ..., xn} are respectively defined as points on the decoded contour, and by connecting all the points on the decoded contour of the image with a curve, The contour figure is decoded. As a result, the coded data is created by utilizing the fact that the contour figure can be simply specified by the distance from the base axis. Therefore, a method that enables efficient coding and decoding of the contour figure and a coding method using the method. The decoding device can be realized.

In addition to the first aspect, the second aspect of the present invention further facilitates the process of calculating the distance from the base axis, and the inside and outside of the contour are represented by true or false for each pixel. Regarding a contour coding method for coding a contour figure, the contour figure is decomposed into k (k is a natural number) partial contour figures, and this decomposition is performed by a logic for each pixel of the k partial contour figures. It is possible to express by calculation, and the straight line selected and arranged around the contour figure for each partial contour figure is the base axis,
The intersections of the n contour lines intersecting the base axis at a predetermined angle and the partial contour figure are calculated, and the distances {x1, x2, ..., xn} between each intersection and the base axis are encoded. , A contour coding method in which a contour coding signal is used together with a signal indicating the configuration of a partial contour graphic. The decoding method is such that the encoded signal is decoded and n
The distances of points {x1, x2, ..., xn} and the configuration of the partial contour figure are derived, and the distances from the base axis are {x1, respectively on n straight lines intersecting the base axis at a predetermined angle. x2, ..., xn} are the decoded points on the contour, the points on the contour are connected by a curve to form a contour, and each pixel of each partial contour figure is true or false. By expressing and calculating a logical operation for each pixel of each partial contour figure corresponding to the same pixel position of the image of the partial contour figure according to the configuration of the derived partial contour figure, and integrating the logical operation results. The outline figure of the entire image is decoded. As a result, the contour figure is decomposed into the partial contour figures, and the information about how to synthesize the partial contour figures to obtain the contour figure is also handled, so that more efficient encoding and decoding of the contour figure can be performed. A method and an encoding / decoding device using the method can be realized.

A third aspect of the present invention is applied to a predictive coding method used for coding moving images and the like, and is a contour coding method for coding a contour figure, and predicts a contour figure to be coded. A contour figure is generated, n feature points are taken on the contour line of the predicted contour figure, and a distance {x to the contour figure at an image position opposite to each of the feature points of the predicted contour figure
This is a contour coding method in which 1, x2, ..., xn} are coded to obtain a contour coded signal. The decoding method is to decode a coded signal to derive a distance {x1, x2, ..., xn} of n points, generate a predicted contour figure of a coded contour figure, and calculate the predicted contour figure. Taking n feature points on the contour line of the figure, the distance {x1, x2, ..., xn} from each of the feature points is taken as a point on the decoded contour, and on all of the decoded contours of the image. By connecting the points of with a curve, the contour figure of the image is decoded. As a result, even in the predictive coding method,
It is possible to realize a method capable of efficiently encoding and decoding a contour figure and a contour encoding and decoding apparatus using the method.

A fourth aspect of the present invention enables the third aspect of the invention to be processed more efficiently, and is a contour coding method for coding a contour figure, wherein a predicted contour figure of a contour figure to be coded is obtained. A straight line selectively generated and arranged around the contour figure is set as a first base axis, and a straight line selectively arranged around the predicted contour figure is set as a second base axis, and a predetermined base line is set as the first base axis. The intersections of the n straight lines intersecting at the angle and the contour figure, and the intersections of the n straight lines intersecting at the predetermined angle with the second base axis and the predicted contour figure, A first distance between the intersection point of the contour figure and the first base axis and a second distance between the intersection point of the predicted contour figure and the second base axis are calculated, and the first and second base axes are calculated. The difference between the first and second distances {x1, x2, ..., xn} is coded to form a contour coded signal. A contour coding method. In this decoding method, the encoded signal is decoded and the distance difference {x1, x
2, ..., xn} is derived to generate a predicted contour figure of the encoded contour figure, and the prediction is performed on n straight lines that intersect the base axis of the predicted contour figure at a predetermined angle. The n points whose distances from the contour figure are {x1, x2, ..., xn} are defined as points on the decoded contour, and all the points on the decoded contour of the image are connected by a curve. By doing so, the contour figure of the image is decoded. As a result, since the contour figure and the predicted contour figure can be processed independently, it is possible to realize a more efficient contour figure coding and decoding method in predictive coding, and a contour coding and decoding apparatus using the method. Is.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION Next, specific examples of the present invention will be described.

(Embodiment 1) FIG. 1 is a conceptual diagram of a contour coding method and a contour decoding method of the first invention.

In the present embodiment, for convenience of description, an example is given in which the contour figure is two-dimensional and divided into two-dimensional blocks before processing. However, the present invention is not limited to this, and the periphery of the contour figure is predetermined. After cutting out to the shape of, the cutout frame may be processed as a reference, or the image frame displaying the contour figure may be used as it is as a reference for processing.

In FIG. 1, in a two-dimensional contour figure image (left in FIG. 1), the filled area indicates an object, and the boundary between the filled portion and the white portion is the contour of the object. A block is obtained by dividing the image frame of the contour figure into 16 parts and cutting out a part thereof. The vertical direction of the block is shown as the X-axis, and the horizontal direction is shown as the Y-axis (Fig. 1, right). The advantage of blocking is that when the contour figure has a complicated shape, most of the contour line can be represented by a low-order polynomial (smooth curve) by cutting out a part of it as a block.

Then, the line is divided into eight straight lines along the bottom of the block, and the distance from the bottom of the block to the contour line {x0,
x1, ..., x7} is calculated. The distance {x0, x1, ..., x7} is DCT
If orthogonal transformation such as is used, encoding can be performed with a small number of bits.

On the other hand, by using {x0, x1, ..., x7} as the X coordinate, it is possible to know the Y-axis position of the straight line indicated by the eight arrows used for the division, so specify it as a point on the contour line. Therefore, if the coordinate points corresponding to {x0, x1, ..., x7} are connected by a smooth curve, the contour can be approximated accurately (in FIG. 1, x0, x1 ,. .., x7 is displayed, but it does not represent the Y-axis coordinate value but merely the length of the straight line indicated by the arrow, and is displayed on the Y-axis for convenience).

From the above viewpoint, if {x0, x1, ..., x7} is coded, the contour can be decoded, and {x0, x
It is clear that 1, ..., x7} can be efficiently coded by orthogonal transform such as DCT.

FIG. 2 is a block diagram of an embodiment of a contour coding apparatus using this method. In the figure, 1 is a contour figure, 2 is a blocker for dividing the contour figure into blocks,
3 is a sampler for dividing the block into n + 1 parts to extract the boundary value, 4 is a contour detector for detecting the contour position at the boundary, 5 is an orthogonal transformer, 6 is an encoder, and 7 is an encoder. It is a signal.

The first embodiment of FIG. 2 configured as described above will be described. It is assumed that the contour figure is represented as an image having a value of true inside the contour and false outside the contour. That is, the filled area in FIG. 1 is true and the white area is false. The blocker 2 divides the contour figure 1 into rectangular blocks and cuts out as shown in the right side of FIG. With one side of the cut-out rectangular block as the base axis, the sampler 3 has n straight lines provided along the base axis (in the block cut out in FIG. 1, seven straight lines with arrows are The block is divided into n + 1 according to (corresponding). For example, if the horizontal line at the lower end of the block is used as the base axis and is divided by vertical lines, n vertical lines in the block will be the dividing boundaries. Contour detector 4
Calculates the number of true values located on the boundary line from the base axis for each boundary line. This value is {x0, x1, ..., x
7}, which represents the distance from the base axis to the contour line. The orthogonal transformer 5 collectively performs orthogonal transformation on {x0, x1, ..., x7}, and the encoder 6 performs variable-length coding on the component to obtain a coded signal. DCT, D for orthogonal transform
Although there are ST, KLT, etc., color signals are often DCT-encoded, and DCT is desirable in consideration of sharing of color signal encoding devices and devices. Further, since the contour line is usually smooth and the orthogonal transform component corresponding to the high frequency component hardly occurs, the number of bits can be reduced by performing the variable length coding using the bias of the occurrence frequency.

Further, when the encoder 6 encodes, the orthogonal transformation component corresponding to the high frequency component represents a minute change in the contour shape, so that it is visually less deteriorated, and as a result, the encoding is performed more than the low frequency component. Since distortion can be tolerated, if this feature is used to roughly quantize and code the orthogonal transform component corresponding to the high frequency component, the number of bits can be further saved.

As described above, an example in which the contour figure can be efficiently coded in block units has been shown. However, even if the contour figure is processed without dividing into blocks, the coding efficiency of the contour figure is improved. Needless to say.

(Second Embodiment) FIG. 3 is a block diagram of a second embodiment of the contour decoding apparatus of the first invention. In the figure, 7 is a coded signal, 10 is a decoder, 11 is an inverse orthogonal transformer, 12 is a contour generator that generates a contour position at a boundary, 13 is an interpolator that interpolates and generates a contour line, and 14 is a block. An inverse blocker that integrates to form a region figure, 1
Reference numeral 5 is a decoded contour figure signal. The contour decoding apparatus in the figure is for correctly decoding the coded signal 7 coded by the contour coding apparatus in FIG.

A second embodiment of FIG. 3 configured as above will be described. The coded signal 7 is decoded by the decoder 10 and decoded into the same signal as that at the input of the encoder 6 of FIG. The inverse orthogonal transformer 11 performs the inverse transformation of the orthogonal transformer 5 of FIG. 2 and outputs {x0, x1, ..., x7}. Contour generator 12
Generates a point on the contour at a position of a distance {x0, x1, ..., x7} from the base axis on the same boundary as the block dividing boundary of FIG. The interpolator 13 connects the points on the contour determined by the contour generator 12 with a straight line or a spline curve to generate a boundary line within the block. The deblocking unit 14 connects the boundaries between the blocks and decodes the contour figure.

In the first embodiment shown in FIG. 2, the input values are expressed as true and false, and FIG. 3 shows an example of outputting the contour line itself. However, the expression by the true and false values and the expression by the contour line are Either one may be used as it can be easily converted. For example,
Instead of outputting the points on the contour by the contour generator 12 in FIG. 3, a value from the base axis to the contour point is generated as true and the other values are false, and the value of the position where true or false is unknown is determined to be true or false. The same effect can be realized by the interpolator 13 that directly derives from a value in the vicinity where false is determined by a predetermined prediction formula.

As described above, the signal coded by the contour coding apparatus of FIG. 2 can be correctly decoded.

(Embodiment 3) FIG. 4 is a conceptual diagram of a concrete example when setting a base axis in the contour coding method and the contour decoding method of the first invention. 4A shows an example in which a boundary line is drawn vertically with the horizontal axis as the base axis, and FIG. 4B shows an example in which a boundary line is drawn horizontally with the vertical axis as the base axis when encoding the contour in the block. Is.

4 (a), {x0, x1, ..., x7} has few non-zero values but no large values, and conversely {x0, x1, ..., x7} of FIG. 4 (b).
0, x1, ..., x7} have many non-zero values but large values.
Which one has the smaller number of bits after encoding depends on the encoding method. For example, in orthogonal transform such as DCT, the number of transform components corresponding to high frequency components is small, and the number of bits is smaller in FIG. 4A. Less. In this way, the number of coding bits changes depending on the selection of the base axis. Therefore, when an appropriate base axis is selected according to the contour, the coding efficiency is improved as compared with the case where a fixed base axis is always selected.

FIG. 5 is a block diagram of a third embodiment of a contour coding apparatus using this contour coding method. In the figure, differences from the block diagram of FIG. 2 are a base selector 20, an encoder 21 and an encoded signal 22 for selecting a base axis.

The third embodiment of FIG. 5 configured as described above will be described. Note that the devices designated by the same numbers as those in the block diagram of FIG. Based on the contours blocked by the blocker 2, the basis selector 20 determines the basis axis used by the sampler 3. It should be noted that in the present embodiment, the processing via the blocker 2 has been described, but it goes without saying that the same processing can be performed on the entire contour figure without dividing into blocks.

As an index for selecting the base axis, a function for predicting the number of bits after encoding from the contour and the base axis is determined, and the base axis having a small function value (it seems that the number of bits is small) may be selected. . Since it is necessary to inform the contour decoding device in order to correctly decode the selected base axis, the distances to the contour points {x0, x1 ,. . . , X7} and coded by the coder 21 at the same time to obtain a coded signal 22.

The encoder 21 is the orthogonal transformer 5 of FIG.
The encoder is a more general-purpose encoder including a combination of the encoder and the encoder 6, and may be an encoder that performs predictive encoding or the like instead of orthogonal transform.

As described above, a plurality (m number) of base axes are prepared in advance, and the base axes can be switched to further improve the coding efficiency.

(Embodiment 4) FIG. 6 is a block diagram of Embodiment 4 of a contour decoding apparatus corresponding to the contour encoder of FIG. In the figure, differences from the block diagram of FIG. 3 are an encoded signal 22, a decoder 23, a basis selector 24, and an interpolator 25. The contour decoding apparatus shown in the figure is for correctly decoding the coded signal 22 coded by the contour coding apparatus shown in FIG.

The fourth embodiment of FIG. 6 configured as above will be described. It should be noted that the devices designated by the same numbers as those in the block diagram of FIG. The encoded signal 22 is decoded by the decoder 23, the distance to the contour point {x0, x1, ..., x7} is output to the contour generator 12, and the code indicating the base axis is output to the base selector 24. To be done. Since the basis selector 24 selects the same basis axis as the basis selector 20 of FIG. 5, the interpolator 25 can correctly decode the points on the contour using the correct basis axis.

As described above, the signal coded by the contour coding apparatus shown in FIG. 5 can be correctly decoded.

(Fifth Embodiment) FIG. 7 shows an example of a plurality of base axes to be selected. The rectangular frame in FIG. 7 represents either the image frame of the contour figure or the block frame when divided into blocks.

As the base axis, in order to efficiently encode contours of various shapes, it is preferable to use a base axis that is similar to the contour shape that can occur frequently. For example, the base axis is 2
If used normally, the horizontal and vertical axes as shown in FIGS. 7A and 7B are efficient. If four base axes are used, use FIG. 7 (a), FIG. 7 (b), FIG. 7 (c), and FIG. 7 (d) in which the horizontal axis and the vertical axis are diagonally aligned. For example, the average value of the difference from the contour line becomes small, and the coding efficiency improves.

(Embodiment 6) FIG. 8 is a conceptual diagram of a contour coding method and a contour decoding method of the second invention. Although n (8 in FIG. 8) straight lines are used for sampling when the contours in the block are encoded, it is natural that the same straight line (boundary line) intersects with the contour lines of the contour figure in plurals. Exists. That is, there are cases where points on a plurality of contours exist on the same boundary line. For example, the filled area A in FIG. 8 has two points on the contour when divided by a vertical line with the horizontal axis as the base axis.

Therefore, taking advantage of the fact that the area A of FIG. 8 can be formed by the exclusive OR of the area B of FIG. 8 and the area C of FIG. 8, the area A of FIG. 8 is not directly encoded. It will be understood that if each of the regions B in FIG. 8 and the region C in FIG. 8 is encoded, the number of contour points on the boundary line becomes one, and the encoding process becomes easy. In this way, the contour figure showing the region A can be divided into the partial contour figure showing the region B and the partial contour figure showing the region C.

FIG. 9 is a block diagram of Embodiment 6 of a contour coding apparatus using this method. In FIG.
The difference from the block diagram of FIG. 2 is a figure decomposer 30 that decomposes contour figures, a multiplexer 31 that multiplexes the codes of each figure, an encoder 32, and an encoded signal 33. Sampling devices 3a, 3b
The operations of the contour detectors 4a and 4b are the same as those of the sampler 3 and the contour detector 4 of FIG.

The sixth embodiment of FIG. 9 configured as described above will be described. Note that the devices designated by the same numbers as those in the block diagram of FIG. The contour figure blocked by the blocker 2 is decomposed into a plurality of partial contour figures by the figure decomposer 30. For example, the above-mentioned area A in FIG. 8 is divided into area B in FIG. 8 and area C in FIG. Sampling and contour detection are performed for each divided area, the contour data of each area is multiplexed, and information necessary for synthesizing the area by the decoding device (method of synthesizing area A from area B and area C) At the same time, it is encoded by the encoder 32 to obtain an encoded signal 33.

As described above, a complex contour figure having a plurality of contour points on the boundary can be efficiently coded.

(Embodiment 7) FIG. 10 is a block diagram of Embodiment 7 of the contour decoding apparatus. In FIG.
The difference from the block diagram of FIG.
4, a separator 35, and a figure synthesizer 36. Contour generator 1
The operations of 2a and 12b and the interpolators 13a and 13b are the same as those of the contour generator 12 and the interpolator 13 of FIG. 3, respectively. The contour decoding apparatus in the figure is for correctly decoding the coded signal 33 coded by the contour coding apparatus in FIG.

The seventh embodiment of FIG. 10 configured as described above will be described. It should be noted that the devices designated by the same numbers as those in the block diagram of FIG. The coded signal 33 is decoded by the decoder 34, and information indicating the distance {x0, x1, ..., x7} to the contour point of each area and the area synthesis method is output. The separator 35 separates the distance to the contour point so as to correspond to each area, and outputs the distance to the contour generators 12a and 12b corresponding to each area. The decoded contour figure corresponding to each area is combined by the figure combiner 36 based on the information indicating the combining method decoded by the decoder 34.

As described above, the signal coded by the contour coding apparatus of FIG. 9 can be correctly decoded.

(Embodiment 8) FIG. 11 is a block diagram of another embodiment 8 of the contour coding apparatus of the second invention.
In the figure, differences from the block diagram of FIG. 9 are encoders 37a and 37b, a multiplexer 38 that multiplexes the code of each figure and synthesis information of the area, and an encoded signal 39.

The eighth embodiment of FIG. 11 configured as described above will be described. It should be noted that the devices designated by the same numbers as those in the block diagram of FIG. In the seventh embodiment shown in FIG. 9, the positions of the decomposed contours of the respective regions are multiplexed and then encoded, but in the eighth embodiment, the encoders 37a and 37b first encode each region, and after encoding, The multiplexer 38 multiplexes the code of each area and the information necessary for combining the areas. Even if the order of the encoder and the multiplexer is exchanged as in the eighth embodiment, the same effect of the present invention can be realized.

(Ninth Embodiment) FIG. 12 is a block diagram of a ninth embodiment of the contour decoding apparatus according to the second invention.
In the figure, the difference from the block diagram of FIG. 10 is a separator 50 and decoders 51a and 51. The contour decoding apparatus in the figure is for correctly decoding the coded signal 33 coded by the contour coding apparatus in FIG.

The ninth embodiment of FIG. 12 configured as described above will be described. It should be noted that the devices designated by the same numbers as those in the block diagram of FIG. In the seventh embodiment shown in FIG. 10, the encoded signal is decoded and then separated, and the contour is decoded for each region. In the ninth embodiment, first, the separator 50 first determines the code and region for each region. Information necessary for synthesizing is separated, and the contour positions are decoded by the decoders 51a and 51b for each area.

As described above, the signal coded by the contour coding apparatus of FIG. 11 can be correctly decoded.

(Embodiment 10) FIG. 13 is a description of a case where a logical product is used as a logical operation of another method of synthesizing a contour figure in the second invention, and the base axis is switched by selecting the base axis. It is a figure.

In FIG. 13, the filled area has a true value, the white area has a false value, and the position where the true value and the false value change is the contour. The area A in FIG. 13 can be expressed as the logical product of the areas B and C in FIG. 13, so that if the areas B and C in FIG.
Can be correctly decoded. At this time, if the region B is encoded with the horizontal axis as the base axis and the region C is encoded with the vertical axis as the base axis, the region B and the region C will be described in the third embodiment rather than the same base axis. As described above, it is possible to further improve the coding efficiency.

(Embodiment 11) FIG. 14 is a diagram showing the advantage of changing the number of divisions when a contour figure is divided by a plurality of straight lines for sampling. In the above description, it was explained that the sampling means divided n + 1, but the value of the division number was not explained.

The case where the contour figure is divided into blocks in advance will be described. If the size of this block is fixed, the number of divisions may be fixed (for example, 9), but if the size of the block changes, the number of divisions will change accordingly. If not changed, the contour shape cannot be faithfully expressed in the large block, and the coding distortion of the large block becomes visually conspicuous.

Therefore, as shown in FIG. 14, the intervals of a plurality of straight lines for sampling are almost the same, so that the number of samples increases according to the size of the block. As described above, if the number of divisions is increased as the size of the block is increased, the distortion of the contour coding is made uniform and can be made visually inconspicuous. Even for blocks of the same size, the length of the contour line in the diagonal direction is longer than the length of the contour line in the horizontal and vertical directions, so depending on the length of the contour line if the processing can be complicated. The number of divisions can be changed to further improve the coding efficiency.

When the number of divisions is changed, it is necessary to multiplex the code indicating the number of divisions into the contour coded signal so that the contour decoding apparatus can correctly obtain the number of divisions of the block. is there.

(Embodiment 12) FIG. 15 is a conceptual diagram of a contour coding method and a contour decoding method of the third invention.
When encoding a contour figure, a contour figure that serves as a reference to the contour figure may be obtained in advance, and is related to the predictive code used for the hierarchical encoding or the encoding of the moving image. For example, in the case of hierarchical coding, it is a rough contour figure,
In the case of a moving image, predictive coding is performed using decoded contour figures that are temporally close to each other.

If there is a reference contour figure and the contour figure can be accurately estimated from the reference contour figure, the encoding efficiency can be improved by encoding the difference with the reference contour figure instead of directly encoding the contour figure. Can be improved. FIG. 15 shows an example of the case where the contour figure is processed into blocks.
FIG. 15A shows a coded block, and FIG. 15B shows a reference block generated from a reference contour figure. The difference between the coded block and the reference block is shown in FIG. 15C, and each arrow represents the difference from a point on each contour of the reference block to the contour of the coded block, and the line of each arrow is drawn. The positions shown are the contour figure of FIG.
The distance between the feature points arranged on the contour line of the reference contour figure in (b) is shown. Point this arrow (upward is positive,
The difference block of FIG. 15 (d) can be constructed by considering the size as {x0, x1, ..., x7} with the sign (downward negative) as the sign and the horizontal axis as the base axis.

The difference block of FIG. 15 (d) can be treated in the same manner as the block-shaped contour figure described above, and {x0, x1, ..., x7} is the distance to the contour point. Should be encoded. At this time, when the reference block and the encoded block are almost the same, {x0, x1, ..., x7} is almost 0, so that encoding is unnecessary and the encoding efficiency can be improved.

FIG. 16 is a block diagram of a twelfth embodiment of a contour coding apparatus using this coding method. In the figure, the difference from the block diagram of FIG. 5 is that the predicted contour figure 40 that is the reference contour figure, the blocker 41, the basis selector 42, the sampler 43, the contour detectors 44 and 45, and the differencer 4
6, an encoder 47, and an encoded signal 48. The operations themselves of the contour detectors 44 and 45 are the same as those of the contour detector 4 in the block diagram of FIG.

The twelfth embodiment of FIG. 16 configured as described above will be described. Since the devices designated by the same numbers as those in the block diagram of FIG. 5 perform the same operations, the description thereof will be omitted. The predicted contour figure 40 is divided into blocks by the blocker 41 and is output as a reference block signal. The basis selector 42 determines the basis axis of sampling from the reference block signal and notifies the samplers 3 and 43. The position of the contour point of the reference block signal is calculated by the sampler 43 and the contour detector 44. The differentiator 46 subtracts the position of the contour point of the reference block from the position of the contour point of the encoded block, and the encoder 47 encodes the difference of the subtractor 46 to encode the encoded signal 4
8 is assumed.

As described above, the contour figure can be efficiently coded by using the reference contour figure which is the predicted contour figure.

(Thirteenth Embodiment) FIG. 17 shows the first embodiment.
23 is a block diagram of Embodiment 13 of the contour decoding apparatus using the method of 2. In the figure, the difference from the block diagram of FIG. 6 is that a predicted contour figure 40 that is a reference contour figure, a blocker 41, a basis selector 42, a sampler 43, and a contour detector 4
4, coded signal 48, decoder 60, and adder 61. The coded signal 33, the decoder 34, the separator 35, and the graphic synthesizer 36. The contour decoding apparatus shown in FIG.
This is for correctly decoding the coded signal 33 coded by the contour coding device 6 of FIG.

The thirteenth embodiment of FIG. 17 configured as above will be described. Since the devices designated by the same numbers as those in the block diagram of FIG. 6 and the block diagram of FIG. 16 perform the same operation, the description thereof will be omitted. The encoded signal 48 is decoded by the decoder 60, and the position of the contour point of the reference block is added by the adder 61. The sum of the adder 61 is the contour generator 12
Is input to the same as that of the fourth embodiment shown in FIG.

Since the predicted contour figure is the same as that of the contour coding apparatus of FIG. 16, the output of the contour detector 44 and the base axis of the base selector 42 are the same as those of the contour coding apparatus of FIG. Since it exactly matches the output of the numbered device,
It can be decrypted correctly.

(Embodiment 14) FIG. 18 is a diagram for explaining one method of selecting a base axis and drawing a plurality of straight lines (boundary lines) for dividing by n + 1 for sampling. .

The case where the contour figure is divided into blocks will be described. In FIG. 18, the base axis is selected so that the curved portion of the contour figure can be specified and the distance to the contour figure is made as short as possible.

The method of dividing the rectangular block into n + 1 along the selected base axis can be arbitrarily performed.
If the horizontal and vertical frames of this rectangular block are the coordinate axes, the position on the boundary is {x0, x if the boundary is diagonal to this coordinate axis.
It is necessary to derive 1, ..., x7} from the two-dimensional coordinates by calculation. However, the boundary is horizontal (Fig. 18 (a))
Alternatively, in the case of vertical (FIG. 18B), the difference calculation in the one-dimensional direction (horizontal or vertical) is performed, and therefore the calculation is easy. Therefore, the processing amount of the present invention can be reduced by dividing the boundary line into a horizontal line or a vertical line.

To determine whether the boundary line is horizontal or vertical, the intersection angle between the base axis and the horizontal axis or the vertical axis is calculated, and the intersection angle close to 90 degrees may be selected. .

(Embodiment 15) Further, the first to fourth inventions are realized by a program, which is recorded on a recording medium such as a floppy disk and transferred, so that it can be easily carried out by another independent computer system. can do. FIG. 19 shows a floppy disk as an example of the recording medium.

In this embodiment, the floppy disk is shown as the recording medium, but an IC card or a CD is used.
-If the program can be recorded, such as a ROM or a cassette, the present invention can be similarly implemented.

[0073]

As described above, the contour figure encoding according to the present invention,
According to the decoding method and the contour encoder and decoder using the method, the contour figure can be efficiently encoded.

Further, by processing this prescription in block units, the same block unit as that of color signal encoding can be obtained, so that it is easy to add a contour figure to the conventional color signal encoding method. . Further, there is a correlation between the color signal and its contour graphic, and if color coding is realized using the information of the contour graphic in block units, the coding efficiency of color coding can be improved.

[Brief description of drawings]

FIG. 1 is a conceptual diagram of a contour coding method and a contour decoding method according to the first invention.

FIG. 2 is a block diagram of the first embodiment of the contour coding apparatus according to the first invention.

FIG. 3 is a block diagram of a second embodiment of the contour decoding apparatus of the first invention.

FIG. 4 is a conceptual diagram of a concrete example of setting a base axis in the contour encoding method and the contour decoding method of the first invention, and FIG. 4A is a vertical boundary with a horizontal axis as a base axis. The figure which shows the example which drew the line (b) is the figure which shows the example which drew the boundary line horizontally with the vertical axis as the base axis.

FIG. 5 is a block diagram of a third embodiment of a contour coding apparatus to which a function of selectively setting a base axis is added in the first invention.

FIG. 6 is a block diagram of a fourth embodiment of a contour decoding apparatus to which a function of selectively setting a base axis is added in the first invention.

FIG. 7 is an explanatory view showing an example of a base axis to be selected in the first invention, (a) showing a horizontal axis, (b) showing a vertical axis (c), (d). Is a diagram showing an example of diagonal axes

FIG. 8 is a conceptual diagram of a contour coding method and a contour decoding method of the second invention.

FIG. 9 is a block diagram of a sixth embodiment of a contour coding apparatus according to the second invention.

FIG. 10 is a seventh embodiment of the contour decoding apparatus of the second invention.
Block diagram of

FIG. 11 is a block diagram of another embodiment 8 of the contour coding apparatus according to the second invention.

FIG. 12 is a block diagram of another embodiment 9 of the contour decoding apparatus according to the second invention.

FIG. 13 is a diagram showing an example of region synthesis using a logical product and a plurality of basis axes in the second invention.

FIG. 14 is a diagram showing the advantage of changing the number of divisions for sampling.

FIG. 15 is a conceptual diagram of a contour coding method and a contour decoding method according to a fourth invention, in which (a) shows a block to be coded, (b) shows a reference block, and (c) shows a difference block. FIG. 6D is a diagram showing a result of processing in which the base axis of the difference block is changed.

FIG. 16 is a first embodiment of a contour coding device according to a fourth invention.
Block diagram of 2

FIG. 17 is a first embodiment of a contour decoding device according to a fourth invention.
Block diagram of 3

FIG. 18 is a diagram showing an example of how to draw a division boundary line for selecting a base axis and sampling.

FIG. 19 is a diagram showing an example of a recording medium on which the method of the present invention is recorded.

[Explanation of symbols]

 2 Blocker 3,43 Sampler 4,44,45 Contour detector 5 Orthogonal transformer 6,21,32,39,47 Encoder 10,23,34,51,60 Decoder 11 Inverse orthogonal transform Device 12 contour generator 13,25 interpolator 14 deblocker 20, 24, 42 basis selector 30 graphic decomposer 31,387 multiplexer 35, 50 separator 36 graphic synthesizer 46 differencer 61 adder

Claims (42)

[Claims] 1. A contour coding method for coding a contour figure, comprising a straight line selectively arranged around the contour figure as a base axis, and n straight lines intersecting the base axis at a predetermined angle. A contour coding method in which intersections with the contour are calculated, and the distances {x1, x2, ..., xn} between the intersections and the base axis are coded to obtain a contour coded signal. 2. A contour coding method for coding a contour figure in which the inside and outside of the contour are expressed as true or false for each pixel, wherein the contour figure is divided into k (k is a natural number) partial contour figures. The decomposition is performed so that the contour figure can be expressed by a logical operation for each pixel of the k partial contour figures, and a straight line selectively arranged around the contour figure for each of the partial contour figures is a base axis. Then, the intersections of n straight lines intersecting the base axis at a predetermined angle and the partial contour figure are calculated, and the distance {x1, x2, ..., xn} between each intersection and the base axis is calculated. A contour coding method which is coded into a contour coded signal together with a signal indicating the configuration of the partial contour figure. 3. The contour coding method according to claim 2, wherein exclusive OR is used as the logical operation. 4. The contour coding method according to claim 2, wherein the contour figure is decomposed into partial contour figures so that the number of intersections with the contours on n straight lines intersecting the base axis at a predetermined angle is at most 1. 5. A contour coding method for coding a contour figure, wherein a predicted contour figure of the contour figure to be coded is generated, and n feature points are taken on the contour line of the predicted contour figure.
A contour coding method for coding a distance {x1, x2, ..., xn} to the contour graphic at an image position opposite to each of the feature points of the predicted contour graphic to obtain a contour coded signal. 6. The contour coding method according to claim 5, wherein the distance in the perpendicular direction of the predicted contour figure at the feature points is {x1, x2, ..., xn}. 7. A contour coding method for coding a contour figure, wherein a predicted contour figure of a contour figure to be coded is generated, and a straight line selectively arranged around the contour figure is used as a first base axis. , A straight line selectively arranged around the predicted contour figure is set as a second base axis, and an intersection point between the n straight lines intersecting the first base axis at a predetermined angle and the contour figure is calculated, and An intersection of n straight lines intersecting the second base axis at a predetermined angle and the predicted contour figure is calculated, and a first distance between the intersection of the contour figure and the first base axis, and A second distance between the intersection of the predicted contour figure and the second base axis is calculated, and the difference {x1, x2, ..., xn} between the first and second distances at the opposite image positions is calculated. A contour coding method for coding to obtain a contour coded signal. 8. The contour coding method according to claim 5, wherein the predicted contour graphic is generated from a temporal graphic that is a contour graphic of a moving image, and is temporally preceding, subsequent, or preceding and following. 9. The base axis divides the contour figure into pixel areas of a predetermined shape, and is either a straight line of a specific side of the pixel area of the predetermined shape or a straight line drawn from the apex of the predetermined shape inside the predetermined shape. 8. The contour coding method according to claim 1, wherein the straight line is determined as a base axis. 10. The contour encoding method according to claim 9, wherein the predetermined shape is a rectangle, a pixel division method in the rectangle shape is different for each rectangular area, and the rectangular area division method is encoded as additional information. . 11. The base axis is set by selecting one from m kinds of base axes, and the contour encoded signal includes a selection signal indicating the selected base axis. 8. The contour coding method according to any one of 2, 5, and 7. 12. The contour coding method according to claim 1, wherein a horizontal line, a vertical line, or a diagonal line is used as a base axis. 13. The contour coding method according to claim 1, wherein the value of n is changed according to the length of the base axis. 14. The contour coding method according to claim 1, wherein a straight line intersecting the base axis is a horizontal line or a vertical line only. 15. A contour decoding method for decoding a coded signal of the contour coding method according to claim 1, wherein the coded signal is decoded to derive {x1, x2, ..., xn}. Then, on the n straight lines intersecting the base axis at a predetermined angle, the n points whose distances from the base axis are {x1, x2, ..., xn} are the points on the decoded contour, A contour decoding method for decoding a contour figure of the image by connecting all points on the decoded contour of the image with a curve. 16. A contour decoding method for decoding a coded signal of the contour coding method according to claim 2, wherein the coded signal is decoded and a distance of n points {x1, x2, ..., xn. } And the configuration of the partial contour figure, and the distance from the base axis is {x1, x2, ..., x on each of n straight lines intersecting the base axis at a predetermined angle.
n points, which are n}, are points on the decoded contour, the points on the contour are connected by a curve to form a contour, and each pixel of each partial contour figure is represented by true or false, The contour of the whole image of the image is calculated by calculating a logical operation for each pixel of each partial contour figure corresponding to the same pixel position of the figure image according to the configuration of the derived partial contour figure, and integrating the results of the respective logical operations. A contour decoding method for decoding a figure. 17. The contour decoding method according to claim 16, wherein exclusive OR is used as the logical operation. 18. A contour decoding method for decoding a coded signal of the contour coding method according to claim 5, wherein the coded signal is decoded and a distance of n points {x1, x2, ..., xn. }, And generates a predicted contour figure of the one-way encoded contour figure,
Taking n characteristic points on the contour line of the predicted contour figure, the distance {x1, x2, ..., xn} from each characteristic point is the decoded point on the contour, and all the decoding of the image is performed. A contour decoding method for decoding a contour figure of the image by connecting points on the contour with a curve. 19. The contour decoding method according to claim 18, wherein the vertical distance of the predicted contour figure at the feature point is {x1, x2, ..., xn}. 20. A contour decoding method for decoding a coded signal of the contour coding method according to claim 7, wherein the coded signal is decoded to obtain a difference in distance {x1, x2, ..., xn. } Is derived,
On the other hand, a predicted contour figure of the coded contour figure is generated, and the distance from the predicted contour figure is {x on each of n straight lines intersecting the base axis of the predicted contour figure at a predetermined angle.
1, x2, ..., xn} are defined as points on the decoded contour, and all the points on the decoded contour of the image are connected by a curve to form the contour figure of the image. A contour decoding method for decoding. 21. The contour decoding method according to claim 18, wherein the predicted contour figure is a contour figure of a moving image, and the predicted contour figure is temporally preceding, subsequent, or preceding and succeeding images. 22. The base axis divides the contour figure into pixel areas of a predetermined shape, and is either a straight line of a specific side of the pixel area of the predetermined shape or a straight line drawn from the apex of the predetermined shape inside the predetermined shape. 21. The contour decoding method according to claim 15, wherein the straight line is determined as a base axis. 23. The rectangular pixel division method is different for each rectangular area, and the rectangular area pixel division method is obtained by decoding the rectangular area division method encoded as additional information. Contour decoding method. 24. The basis axis is set by selecting one of the m number of basis axes, and the contour coded signal includes a selection signal indicating the selected basis axis. 21. The contour decoding method according to claim 16, 18, or 20. 25. The contour decoding method according to claim 15, 16, 18 or 20, wherein a horizontal line, a vertical line or a diagonal line is used as a base axis. 26. The contour decoding method according to claim 15, 16, 18, or 20, wherein the value of n is changed according to the length of the base axis. 27. The contour decoding method according to claim 15, wherein only a horizontal line or a vertical line intersects the base axis. 28. A recording medium for a computer, in which a program for realizing at least one of claims 1, 2, 5, 7, 15, 16, 18 or 20 is recorded. 29. An encoding device for encoding a contour figure in which the inside and the outside of the contour are expressed as true or false for each pixel, wherein the basis selecting means selects a basis axis for sampling the contour figure. Sampling means that outputs the true and false values of the division boundary by dividing n + 1 along the base axis selected by the base selecting means, and the true and false values of the output of the sampling means from true to true for each division position The contour detecting means for detecting and outputting a false position or a position changing from false to true, and an encoding means for encoding a code indicating the output of the contour detecting means and the base axis selected by the base selecting means are provided. A contour coding device in which the output of the coding means is a contour coded signal. 30. A decoding device for decoding a contour figure by decoding a contour coded signal obtained by coding a contour position of a contour figure and a code indicating a base axis for specifying the contour position. Decoding means for decoding the contoured signal to output a contour position and a base selection signal, and a contour for generating a true / false value indicating whether it is inside the contour or outside the contour according to the contour position of the decoding means Generation means, basis selection means for selecting a sampling basis axis from the basis selection signal of the decoding means, and true / false indicating the inside and outside of the contour by arranging the outputs of the contour generation means in line with the basis axis selected by the basis selection means And a contour decoding device for generating a contour figure by integrating the outputs of the interpolation means. 31. An encoding device for encoding a contour figure in which the inside and the outside of the contour are expressed as true or false for each pixel, and a block forming means for dividing the contour figure into blocks of a predetermined shape and outputting the blocks. Sampling means for dividing the predetermined shape block output from the blocking means into n + 1 along a predetermined direction to output a true / false value of the division boundary, and true of the sampling means output for each division position. Contour detection means for detecting and outputting a position where a false value changes from true to false or false to true, and coding means for coding the output of the contour detection means are provided, and the output of the coding means is contour coded. A contour coding device as a signal. 32. A decoding device for decoding a contour figure by decoding a contour coded signal obtained by coding a contour position of a contour figure and a code indicating blocking for specifying the contour position, the contour code being a contour code. Decoding means for decoding the encrypted signal,
Contour generating means for generating a true / false value indicating whether it is inside the contour or outside the contour according to the component of the decoding means;
An interpolating means for arranging the outputs of the contour generating means to generate a true / false value indicating the inside / outside of the contour of the predetermined shape block by interpolation; A contour decoding device for converting the output of the inverse blocking means into a decoded contour figure. 33. An encoding device for encoding a contour figure in which the inside and the outside of the contour are expressed as true or false for each pixel, and a blocking means for dividing the contour figure into blocks of a predetermined shape and outputting the blocks. Basis selecting means for selecting a sampling basis axis from the blocking means output, and n + 1 along the basis axis selected by the blocking means output by the basis selecting means.
Sampling means for dividing and outputting the true and false values of the dividing boundary,
Contour detection means for detecting and outputting the position where the true / false value of the sampling means output changes from true to false or from false to true for each division position, and selected by the contour detection means output and the base selection means A contour coding device comprising coding means for coding a code indicating the base axis, and using the output of the coding means as a contour coded signal. 34. A decoding device for decoding a contour graphic by decoding a contour coded signal obtained by coding a contour position of a contour graphic and a block for specifying the contour position and a code indicating a base axis in the block. And a decoding means for decoding the contour coded signal to output a contour position and a base selection signal, and true / false indicating whether the contour is inside or outside the contour according to the contour position of the decoding means. A contour generating means for generating a value of, a base selecting means for selecting a base axis for sampling from a base selecting signal of the decoding means, and an output of the contour generating means according to the base axis selected by the base selecting means. The interpolating means for generating a true / false value indicating the inside and outside of the outline of the rectangular block by interpolation and the deblocking means for integrating the output of the interpolating means to form the entire figure are provided, and the deblocking means output A contour decoding device for decoding a contour figure. 35. An encoding device for encoding a contour figure in which the inside and the outside of the contour are expressed as true or false for each pixel, wherein the contour figure can be synthesized by logical operation k (k is a natural number).
Figure decomposing means for decomposing into individual partial contour figure blocks, and each partial contour figure block decomposed by the figure decomposing means is divided into n + 1 along a predetermined direction to output a true / false value of a division boundary. Sampling means, contour detecting means for detecting and outputting a position where the true / false value of the sampling means output changes from true to false or from false to true for each division position, and A contour coding apparatus, comprising: coding means for coding a code indicating a method for synthesizing the contour detecting means output and the partial contour figure, wherein the coding means output is a contour coded signal. 36. A decoding device for decoding a contour figure by decoding a contour coded signal obtained by coding a contour position of each partial contour figure of a contour figure and a code indicating a method of synthesizing the partial contour figure. Decoding means for decoding the contour encoded signal and outputting the contour position and the partial contour figure synthesizing method, and a separating means for separating the signal of the contour position output from the decoding means corresponding to each partial contour figure, Contour generating means for generating a true / false value for each partial contour figure indicating whether the contour is inside or outside the contour according to the contour position output from the separating means; and the contour generating means of the contour generating means by a predetermined method. Interpolating means for arranging the outputs to generate a true / false value indicating the inside and outside of the contour of the partial contour figure by interpolation, and a figure synthesizing means for synthesizing the output of the interpolating means of each partial contour figure by the partial contour figure synthesizing method of the decoding means. And the figure A contour decoding device that integrates the outputs of the synthesizing means to form the whole figure to obtain a decoded contour figure. 37. A coding device for coding a contour figure in which the inside and outside of the contour are expressed as true or false for each pixel, and a block forming means for dividing the contour figure into blocks of a predetermined shape and outputting the blocks. A graphic decomposing means for decomposing the block forming means into k (k is a natural number) blocks of partial contour figures which can be combined by a logical operation, and a block of each partial contour figure decomposed by the figure decomposing means in a predetermined direction. Sampling means that outputs the true / false value of the division boundary by dividing n + 1 along the position, and the position where the true / false value of the output of the sampling means changes from true to false or from false to true for each division position And a coding means for coding a code indicating a method for combining the contour detection means output of each partial contour figure and the partial contour figure, and outputting the coding means output as a contour code. A contour coding apparatus for converting the signal into a signal. 38. A contour coded signal obtained by decoding a contour position of each partial contour figure of the contour figure and a block for specifying the contour position and a code indicating a method of synthesizing the partial contour figure is decoded and contoured. A decoding device for decoding a figure, comprising decoding means for decoding a contour coded signal to output a contour position and a partial contour figure synthesizing method, and a contour position signal output from the decoding means for each partial contour figure. And a contour generation unit that generates, for each partial contour figure, a true / false value indicating whether it is inside the contour or outside the contour according to the contour position output from the separating unit. An interpolation means for arranging outputs of the contour generating means by a predetermined method to generate a true / false value indicating the inside and outside of the contour of the predetermined shape block of the partial contour figure by interpolation, and the output of the interpolating means of each partial contour figure Recovery The contour synthesizing means for synthesizing by the partial contour graphic synthesizing method of the encoding means, and the deblocking means for integrating the output of the graphic synthesizing means to form the whole figure, and the contour obtained by decoding the output of the deblocking means A contour decoding device for a figure. 39. An encoding apparatus for encoding a contour figure in which the inside and the outside of the contour are represented by true or false for each pixel by referring to a predicted contour figure generated by prediction, First basis selecting means for selecting a basis axis, and first sampling means for outputting a true / false value of a division boundary by performing n + 1 division along the basis axis selected by the first basis selecting means. When,
First contour detecting means for detecting and outputting a position at which the true / false value of the sampling means output changes from true to false or from false to true for each divided position of the first sampling means; Second basis selecting means for selecting a basis axis for sampling the predicted contour figure, and n + 1 division along the basis axis selected by the second basis selecting means to output a true / false value of the division boundary. And a position at which the true / false value of the output of the sampling means changes from true to false or from false to true for each division boundary of the second sampling means. Second contour detecting means, subtracting means for subtracting the changing position of the second contour detecting means from the changing position of the first contour detecting means on the same division boundary, and a code for coding the difference between the subtracting means. A contour coding apparatus comprising a coding means, wherein the output of the coding means is a contour coded signal. 40. A contour coded signal obtained by coding a difference between contour positions of a predicted contour figure and a contour figure and a code indicating a base axis for specifying the difference is true or false for each pixel inside and outside the contour. A decoding device for decoding a contour figure by decoding with reference to the predicted contour figure represented by, and a base selecting means for selecting a base axis of sampling, and a base axis selected by the base selecting means. Along with the sampling means for dividing by n + 1 and outputting the true or false value of the division boundary, and the true or false value of the output of the sampling means for each division position of the sampling means from true to false or false. Contour detection means for detecting and outputting a position that changes to true, a decoding means for decoding the contour coded signal and outputting a contour position, and the contour detection for a contour position signal output from the decoding means. Adding means for adding the changing positions of the same division position of the means , The interpolation means for arranging the sums of the adding means by a predetermined method to generate a true / false value indicating the inside and outside of the contour by interpolation, and the output of the interpolation means are integrated to form a decoded contour figure forming the entire figure. Contour decoding device. 41. An encoding device for encoding a contour figure in which the inside and outside of the contour are expressed as true or false for each pixel with reference to a predicted contour figure generated by prediction, wherein the contour figure has a predetermined shape block. First block forming means for dividing and outputting the predicted contour figure into pixels at the same position as the first block forming means, and the second block Basis selecting means for selecting a base axis for sampling from the output of the converting means, and first for outputting the true value of the division boundary by dividing the output of the first blocking means by n + 1 along the selected base axis. And the first sampling means for detecting the position where the true / false value of the sampling means output changes from true to false or from false to true for each divided position of the first sampling means. The contour detection means and the output of the second blocking means Second sampling means for dividing the boundary by n + 1 along the selected base axis and outputting a true / false value of the division boundary, and the sampling means output for each division position of the second sampling means. Second contour detecting means for detecting and outputting a position at which the true / false value of changes from true to false or from false to true, and the second contour detecting means from the changing position of the first contour detecting means at the same division boundary. A contour coding apparatus comprising: a subtracting means for subtracting the changed position of the contour detecting means; and a coding means for coding the difference of the subtracting means, wherein the output of the coding means is a contour coded signal. 42. A contour coded signal obtained by coding a difference between the contour position of a predicted contour figure and a contour position of the contour figure and a code for indicating the difference and blocking and a base axis is true or false for each pixel inside and outside the contour. A decoding device for decoding a contour figure by referring to a predicted contour figure represented by false and decoding the contour figure, the block forming means dividing the predicted contour figure into blocks of a predetermined shape, and outputting the blocks. Basis selecting means for selecting a sampling basis axis from the means output, and sampling means for outputting the true or false value of the division boundary by dividing the blocking means output by n + 1 along the selected basis axis. A contour detecting means for detecting and outputting a position where the true / false value of the sampling means output changes from true to false or from false to true for each division position of the sampling means; Decode to output contour position Conversion means, addition means for adding the change position of the same division position of the contour detection means to the contour position signal output from the decoding means, and the sum of the addition means is arranged by a predetermined method to arrange inside and outside of the contour of the rectangular block. To form a whole figure by interpolating an interpolating means for generating a true / false value indicating that, a graphic synthesizing means for synthesizing the output of the interpolating means by the graphic synthesizing method of the decoding means, and an output of the graphic synthesizing means. A contour decoding apparatus which comprises a deblocking means for converting the output of the deblocking means into a decoded contour figure.