JPH0879759A - Image coder and image decoder - Google Patents

Abstract

PURPOSE: To generate an excellent predicted image by matching a grating point on a predicted reference image with a moving contour part in a prediction coding system where an image is divided into rectangular blocks by a rectangular grating and a moving vector at the grating point is used to convert the shape of each rectangular block. CONSTITUTION: A moving contour detection section 1 detects a moving contour based on a reference image and other decoded image. A grating point determination section 2 sets grating points in matching the moving contour obtained by the moving contour detection section 1. A motion estimate section 3 estimates to which on a coded object image the grating points on a referenced image obtained by the grating point determination section 2 are moved. A prediction section 4 obtains a predicted image by using the reference image and the grating point moving vector on the reference image. Thus, the moving contour detection section 1 obtains the moving contour and the grating points used for the prediction match the moving contour, then the efficiency for estimating the motion is improved and an excellent predicted image is generated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus and an image decoding apparatus, and more particularly, it divides an image into rectangular blocks by a rectangular grid and uses the motion vector of the grid points to form the shape of the rectangular block. The present invention relates to an image coding apparatus and an image decoding apparatus that create a good predicted image by matching a grid point on a predicted reference image with a moving contour in a predictive coding method that performs conversion. For example, it is applied to high-efficiency encoding of image data in the field of digital image processing.

[0002]

2. Description of the Related Art Conventionally, there has been proposed a method for highly efficient encoding of a digital moving image by using techniques such as inter-frame prediction, orthogonal transformation, quantization, variable length encoding and the like. For example, the publicly known document "MPEG2 Interframe Prediction Method" (Technical Report of the Television Society, Vol. 16, No 61. pp. 37-42, 19).
There is a method described in October 15, 1992).

As a method of interframe prediction, a prediction method using affine transformation has been studied. For example,
Known document "Very Low Bitrate Video Coder usin
g Warping Prediction ”(1993 Image Coding Symposium 8-7, pp.167-168) and the known document“ A Novel Vid ”.
eo Coding Scheme Based on Temporal Prediction
Using Digital Image Warping "(IEEE Inter
national Conferenceon Consumer Electronics, 199
In the method described in 3), as shown in FIG. 6, the motion of the image is expressed by a modified grid, and the reference image (usually,
In the encoding device, the image is locally decoded and stored in the frame memory, and in the decoding device, it is the image already decoded and stored in the frame memory). Here, ● represents a grid point.

FIG. 7 is a block diagram of a predictive image creating circuit in a conventional image coding apparatus. In the figure, 21 is a contour detecting section,
Reference numeral 22 is a grid point determination unit, 23 is a motion estimation unit, and 24 is a prediction unit. The contour detector 21 detects the contour of the reference image. For contour detection, for example, there is a method using a primary difference or a gradient operator, a method using a Sobel operator, or the like. These are described in detail in the publicly known document "Computer Vision" (edited by Taniuchida, Maruzen Co., Ltd. (1990)).

The grid point determining unit 22 sets grid points on the reference image in accordance with the contour obtained by the contour detecting unit 21.
For example, the initial grid is a square grid of 16 pixels × 16 pixels, and each grid point is moved to the nearest contour. However, a grid that can be created after moving
Call each side) so that they do not overlap. In addition, the grid point should not be moved too far to the nearest point.
That is, the range in which the grid points can move from the initial grid points is limited. An example of the modified grid on the reference image created in this way is shown in FIG.

In the motion estimation unit 23, the grid point determination unit 2
It is estimated to which position on the encoding target image the grid point on the reference image obtained in 2 has moved. As a method of motion estimation, a method is used in which an area made up of grid points on the reference image and pixels in the vicinity thereof is taken and which area on the image to be encoded matches this area (hereinafter referred to as “block matching”). Call) is used. Specifically, consider an area of M pixels × N pixels centered on a grid point on the reference image, check the degree of coincidence with an area of the same size on the image to be encoded, and determine the center of the area with the best degree of coincidence. Is the destination of the grid point. A vector representing the movement at this time is called a motion vector. As the degree of coincidence between regions, the sum of absolute value errors of pixel values in the region or the weighted sum of absolute value errors is used.

The prediction unit 24 obtains a predicted image using the lattice points on the reference image, the motion vector and the reference image. As a method of obtaining a predicted image, there are methods such as affine transformation using a triangular block and motion vector interpolation in a rectangular block. Regarding these, the publicly known document “Very Low Bitrate Video Coder using Wa” is mentioned.
rping Prediction ”. The difference data between the prediction image obtained by the prediction unit 24 and the image to be encoded is encoded by an encoding unit (not shown), incorporated into the encoded data, and transmitted or accumulated. or,
The encoded data is decoded by a decoding unit (not shown) and added to the predicted image to obtain a decoded image. The decoded image is stored in a frame memory (not shown) and used as a reference image in the subsequent encoding.

FIG. 9 is a block diagram of a predictive image forming circuit in a conventional image decoding apparatus.
2 is a grid point determination unit, 33 is a motion vector decoding unit, and 34 is a prediction unit. Since the contour detection unit 31, the grid point determination unit 32, and the prediction unit 34 operate in the same manner as the encoding device in FIG. 7, their description will be omitted. The motion vector decoding unit 33 is a unit that extracts and decodes the code of the motion vector incorporated in the encoded data. The output is the decoded motion vector and is input to the prediction unit 34.

Since the decoding apparatus obtains the same reference image as that of the encoding apparatus, the contour detecting section 31 and the lattice point determining section 3
The same as the modified grid point obtained by the encoding device is obtained by 2. The motion vector obtained by the motion vector decoding unit 33 is also the same as that obtained by the encoding device. Therefore, the prediction image obtained by the prediction unit 34 is the same as that obtained by the encoding device. The predicted image obtained by the prediction unit 34 and the decoded value of the difference data obtained by the decoding unit (not shown) are added together to obtain a decoded image. The decoded image is displayed on a display or the like and is also stored in a frame memory (not shown) to be used as a reference image in the subsequent decoding.

[0010]

As described above, in the conventional image coding apparatus and image decoding apparatus, when determining the grid points on the reference image, the contour on the reference image is obtained and the contour is matched with the contour. To set grid points. However, some of the obtained contours do not actually move, and it is inefficient to match the grid points with these stationary contours.
For example, when a stationary contour and a moving contour exist near the initial grid point and the stationary contour is closer, the grid point moves to the stationary contour according to the conventional method. For this reason, the motion estimation unit may not be able to estimate the motion of a moving contour, and the prediction efficiency may decrease.

Particularly in a human face image and the like, information concentrates on the movement of the eyes and mouth, but at such a time, it is inconvenient to set the lattice points on the contour portion of the face that does not move. Further, even in the case of images other than face images, when looking at an image, attention is paid to the place where there is motion due to the visual characteristics, so it seems that the overall image quality deteriorates if the prediction efficiency decreases in the moving part. In the conventional method, there is a problem in that the motion estimation unit may not be able to estimate the motion of a moving contour, thus reducing the prediction efficiency.

The present invention has been made in view of the above circumstances, and predictive coding is performed in which an image is divided into quadrilateral blocks by a quadrilateral lattice and the motion vector of the lattice points is used to perform the shape conversion of the quadrilateral block. In the method, it is an object to provide an image encoding device and an image decoding device that create a good prediction image by matching the lattice points on the prediction reference image with the moving limbus.

[0013]

In order to solve the above-mentioned problems, the present invention (1) divides an image into rectangular blocks by a rectangular grid, and uses a motion vector of the grid points to transform the shape of the rectangular block. In the predictive image generation circuit in the moving image predictive coding apparatus that uses a predictive image, a moving contour is detected using a reference image used for prediction and a decoded image of at least one frame other than the reference image that has already been processed. The moving contour detecting means, the grid point determining means for setting the grid points on the reference image in accordance with the contour obtained by the moving contour detecting means, and the grid point on the reference image obtained by the grid point determining means. A motion estimation means for estimating a position on the target image to be encoded, and a prediction means for obtaining a predicted image using the lattice points on the reference image, the motion vector and the reference image are provided. And (2) moving area detecting means for detecting a moving area using a reference image used for prediction and a decoded image of at least one frame other than the reference image which has already been processed. A contour detecting means for detecting a contour on a reference image used for prediction, and a moving contour selecting means for selecting a moving contour, and (3) the moving area detection in (2). The means includes a difference means for obtaining a difference between a reference image used for prediction and a decoded image other than the reference image, an absolute value calculating means for obtaining an absolute value of difference data obtained by the difference means, and the absolute means. Area detection means for detecting an area by threshold processing from the absolute value by the value calculation means, and (4) in (1), the moving contour detection means used for prediction. A motion estimation means for estimating a motion of the contour by using the previously processed decoded image over at least one frame other than the reference image,
Or a moving contour selecting means for selecting a moving contour, or (5) the image is divided into rectangular blocks by a rectangular grid, and the shape of the rectangular block is deformed using the motion vector of the grid points, In the predictive image generation circuit in the moving image predictive decoding device as a predictive image,
Reference image and at least one other than the reference image used for prediction
Moving contour detecting means for detecting moving contours using already processed decoded images of frames or more, and grid point determination for setting grid points on the reference image in accordance with the contours obtained by the moving contour detecting means Means, a motion vector decoding means for taking out and decoding a motion vector incorporated in the encoded data, and a prediction means for obtaining a prediction image using the lattice points on the reference image, the motion vector and the reference image. thing,
Further, (6) in (5) above, the moving contour detecting means detects a moving region using a reference image used for prediction and a decoded image of at least one frame other than the reference image which has already been processed. And a contour detecting means for detecting a contour on a reference image used for prediction, and a contour selecting means for selecting a moving contour, and (7) in (6) above. A moving area detecting means, a difference means for calculating a difference between a reference image used for prediction and a decoded image other than the reference image, and an absolute value calculating means for calculating an absolute value of difference data obtained by the difference means. A region detecting means for detecting a region by threshold processing from the absolute value by the absolute value calculating means, and (8) in (5), the moving contour detecting means is used for prediction. A motion estimation means for estimating the movement of the contour using a reference image and a decoded image of at least one frame other than the reference image, and a moving contour selection means for selecting a contour having motion. Is.

[0014]

According to the encoding device configured as described above,
A moving contour is detected using a reference image used for prediction and a decoded image of at least one frame that is more than that, and grid points are set according to the moving contour, and the movement of the set grid points is detected. Since it is estimated, the motion of the image can be represented more faithfully. When the moving contour is detected, the moving area detection and the contour detection are performed independently, and the moving contour is obtained from these results, so that the moving contour can be easily obtained. In the detection of the moving area, the absolute value of the difference between the reference image used for prediction and the decoded image that has already been processed is taken, and the moving area is detected by the threshold processing, so that the moving area can be easily obtained. it can. Further, when detecting a moving contour, the contour on the reference image used for prediction is detected, and the movement of the contour is estimated using the reference image used for prediction and the decoded image of at least one frame other than that. And this will select a moving contour,
The moving contour can be accurately obtained.

Further, according to the decoding apparatus configured as described above, a moving contour is detected using the reference image used for prediction and the decoded image of at least one frame other than that, and the grid is detected. Since the points are set according to the moving contour, the same grid as that obtained by the encoding device is required. When detecting a moving contour, moving area detection and contour detection are performed independently, and the moving contour is obtained from these results. Therefore, the moving contour can be easily obtained. The same grid as the one is required.

In the detection of the moving area, the absolute value of the difference between the reference image used for prediction and the other decoded image that has already been processed is obtained, and the moving area is detected by the threshold processing. It can be found and is used to find the same grid as that obtained in the encoder. Further, when detecting a moving contour, the contour on the reference image used for prediction is detected, and the movement of the contour is estimated using the reference image used for prediction and the decoded image of at least one frame other than that. Then, by selecting a contour with movement by this, it is possible to accurately determine the movement territory,
This is used to find the same grid as that obtained with the encoder.

[0017]

Embodiments will be described below with reference to the drawings. FIG. 1 is a block diagram for explaining an embodiment (Claim 1) of an image coding apparatus according to the present invention. In the figure, 1 is a moving contour detection unit, 2 is a grid point determination unit, and 3 is a motion estimation unit. 4 is a prediction unit. The motion estimation unit 3 and the prediction unit 4 have the same functions as those described in FIG. 7. The moving contour detection unit 1 is a portion that detects a moving contour from the reference image and the decoded images other than the reference image by the method of the present invention. Assuming that the frame number of the frame currently being encoded is n, the decoded image of frame number n-1 is used as the reference image for motion compensation prediction. The decoded image of frame number n-2 is used as the decoded image other than the reference image. Here, it is assumed that the frame number is smaller as the frame is older in time.

Since the purpose is to improve the prediction efficiency by setting the grid points according to the contour with motion and estimating the motion, the motion contour is detected using the image to be coded and the reference image. It is also possible. However, in general, when lossy encoding is used, the decoding target image itself cannot be obtained by the decoding device. In that case, in order for the decoding device to obtain the same lattice points as the encoding device, it is necessary to encode the information of the modified lattice on the reference image and incorporate it into the encoded data. Of course, it is possible to improve the prediction efficiency by configuring the present invention in this way.

In the embodiment shown in FIG. 1, in order to eliminate the need for encoding the grid point information of the reference image described above, the moving contour is obtained using the reference image and the other already decoded image. ing. In most cases, the motion in a moving image can be regarded as temporally continuous, so even if the moving contour is detected without using the image to be coded, it is possible to obtain something close to the actual moving contour. Is.

The grid point determination unit 2 sets grid points according to the moving contours obtained by the moving contour detection unit 1. For example, the initial grid is a square grid of 16 pixels × 16 pixels, and each grid point is moved to the nearest moving contour. However, it should not be overlapped with each side of the deformed lattice formed after the movement, and the lattice point should not be moved too far to the nearest point. That is, the range in which the grid points can move from the initial grid points is limited. As described above, in the encoding device, the moving contour is obtained, and the grid points are set according to the moving contour, whereby the prediction efficiency is improved, and thus the coding efficiency can be improved.

FIG. 2 shows the structure of the moving contour detecting section in FIG. 1 (claim 2). In the figure, 5 is a moving area detecting section, 6 is a contour detecting section, and 7 is a moving contour selecting section. The contour detector 6 is shown in FIG.
Works the same as described in. The moving area detection unit 5 detects a moving area from the reference image and the decoded images that have already been processed. Information on the detected moving area is input to the moving contour selecting unit 7. The moving contour selecting unit 7 detects a moving contour from the contour on the reference image detected by the contour detecting unit 6 and the moving region detected by the moving region detecting unit 5. That is, of the contour points, the one included in the moving area is determined as the moving area, and the moving contour is detected. As described above, when the moving contour is detected, the moving area detection and the contour detection are performed independently, and the moving contour is obtained from these results, so that the moving contour can be obtained easily.

FIG. 3 is a block diagram (claim 3) of the moving area detecting section in FIG. 2, in which 8 is an absolute value calculating section and 9 is an area detecting section. The absolute value calculation unit 8 calculates the absolute value of the difference data between the reference image and the other decoded images that have already been processed. The calculated absolute value is output to the area detection unit 9. The area detection unit 9 compares the absolute value calculated by the absolute value calculation unit 8 with a predetermined threshold value, and determines a point where the absolute value is equal to or greater than the threshold value as a moving area. The detected moving area is input to the moving contour selecting unit 7. As described above, in the detection of the moving area, the absolute value of the difference between the reference image used for prediction and the other decoded image that has already been processed is taken, and the moving area is detected by the threshold processing, so that it is easy. The moving area can be obtained.

FIG. 4 is another configuration diagram (claim 4) of the moving contour detecting unit in FIG. 1, in which 10 is a motion estimating unit and 15 is a motion estimating unit.
Is a moving contour selection unit, and other parts having the same operations as in FIG. 2 are denoted by the same reference numerals. The contour detection unit 6 has the same function as that described with reference to FIG. The motion estimation unit 10 is a unit that estimates the motion of the contour point detected by the contour detection unit 6. The motion estimation method uses the same block matching as that performed by the motion estimation unit 3 in FIG. 1 at each contour point.

That is, considering a region of M pixels × N pixels centered on the contour point on the reference image, the degree of coincidence with the region of the same size on the decoded image other than the reference image is examined, and the highest degree of coincidence is obtained. The center of the area is the destination of the contour point. The movement vector at this time becomes the motion vector of the contour point. As the degree of coincidence between the two areas, a sum of absolute value errors of pixel values in the areas, a weighted sum of absolute value errors, or the like is used. Moving contour selection unit 15
Compares the magnitude of the motion vector of each contour point with a predetermined threshold value, and when the magnitude of the motion vector is equal to or larger than the threshold value, the contour point is determined as a motion contour and selected. As described above, when detecting a moving contour, the contour on the reference image used for prediction is detected, and the motion of the contour point is estimated using the reference image used for prediction and other decoded images, and Since a certain contour is selected, the moving contour can be accurately obtained.

Next, the image decoding apparatus of the present invention will be described. FIG. 5 is a block diagram for explaining an embodiment (Claim 5) of the image decoding apparatus according to the present invention. In the figure, 11 is a moving contour detecting unit, 12 is a lattice point determining unit, and 13 is a moving vector decoding. The part 14 is a prediction part. The motion vector decoding unit 13 and the prediction unit 14 have the same functions as those described in FIG. The moving contour detection unit 11 is a portion that detects a moving contour from the reference image and the decoded image other than the reference image by the method of the present invention. This operation is the same as that of the moving contour detecting unit 1 of the corresponding encoding device. As a result, the same moving contour obtained by the encoding device is obtained by the decoding device.

In the grid point determination unit 12, the moving contour detection unit 11
The grid points are set according to the moving contour obtained by.
For example, the initial grid is a square grid of 16 pixels × 16 pixels, and each grid point is moved to the nearest moving contour.
However, each side of the deformed grid formed after the movement is prevented from overlapping with each other, and the grid point is not moved to the nearest point far away. That is, the range in which the grid points can move from the initial grid points is limited. As described above, in the decoding device, the moving contour can be obtained and the grid points can be set according to the moving contour. Also, the modified lattice points obtained by the decoding device are the same as those obtained by the encoding device.

Next, an embodiment described in claim 6 will be described. The moving contour detecting unit 11 of FIG. 5 can be obtained by the configuration shown in FIG. That is, when the moving contour is detected, the moving area detection and the contour detection are performed independently, and the moving contour is obtained based on these results, so that the moving contour can be obtained easily. Further, since the modified grid points are determined according to this moving contour, the modified grid points obtained by the decoding device are the same as those obtained by the encoding device.

Next, an embodiment described in claim 7 will be described. The moving contour detection unit 11 of FIG. 5 has the configuration of FIG.
In addition, it can be obtained by configuring the moving area detection unit 5 of FIG. 2 with the configuration of FIG. That is, in detecting the moving area,
Since the absolute value of the difference between the reference image used for prediction and the other decoded image is taken and the moving region is detected by the threshold processing, the moving region can be easily obtained. Further, since the moving contour is selected using this moving region and the modified grid point is determined according to the moving contour, the modified grid point obtained by the decoding device is the same as that obtained by the encoding device.

Next, an embodiment described in claim 8 will be described. The moving contour detecting unit 11 of FIG. 5 can be obtained by using the configuration shown in FIG. That is, when detecting a moving contour, the contour on the reference image used for prediction is detected, the movement of the contour point is estimated using the reference image used for prediction and the decoded image other than that, and the contour with motion is determined by this. Since the selection is made, the moving contour can be accurately obtained. Further, since the modified grid points are determined according to this moving contour, the modified grid points obtained by the decoding device are the same as those obtained by the encoding device.

The threshold values used in this embodiment are values theoretically or empirically determined, and these may be determined in advance, or may be manually or automatically changed as appropriate. Good. As described above, the same modified grid point as that created by the coding apparatus is obtained, and the predicted image is created using this, so that the same predicted image as that obtained by the coding apparatus can be obtained.

[0031]

As is apparent from the above description, the present invention has the following effects. (1) Effect corresponding to claim 1: In the image encoding device of the present invention, a moving contour is detected using a reference image used for prediction and a decoded image of at least one frame other than that. Since the grid points are set according to the moving contour and the movement of the set grid points is estimated, the movement of the image can be represented more faithfully. In particular, in a human face image or the like, the information concentrates on the movements of the eyes and mouth, but even in such a case, the grid points can be set in the moving contour portion. Further, even when an image other than a face image is viewed, attention is paid to a place where there is a motion due to visual characteristics. However, according to the present invention, grid points are set on the contour of a moving part so that the image quality is improved as a whole. it can. (2) Effects corresponding to claims 2 and 3: When detecting a moving contour, the coding apparatus of the present invention independently performs moving area detection and contour detection, and obtains the moving contour based on these results. The moving contour can be easily obtained, and in detecting the moving area, the absolute value of the difference between the reference image used for prediction and the decoded image other than that is taken, and the moving area is detected by the threshold processing, The moving area can be easily obtained. (3) Effect corresponding to claim 4: Further, when detecting a moving contour, a contour on a reference image used for prediction is detected, and a reference image used for prediction and at least one frame other than that are already processed. The motion of the contour is estimated using the decoded image, and the contour with motion is selected by this,
The moving contour can be accurately obtained. (4) Effect corresponding to claim 5: Further, in the image decoding device of the present invention, a moving contour is detected using a reference image used for prediction and a decoded image of at least one frame other than that. However, since the grid points are set according to the moving contour, the same grid as that obtained by the encoding device can be obtained. (5) Effects corresponding to claims 6 and 7: When the moving contour is also detected in the image decoding device, the moving area detection and the contour detection are performed independently, and the moving contour is obtained based on these results, so that the moving movement can be performed easily. The contour can be determined, which is used to determine the same grid as obtained by the coding device. Further, in the detection of the moving area, the absolute value of the difference between the reference image used for prediction and the decoded image that has already been processed other than that is taken, and the moving area is detected by the threshold processing, so that the moving area is easily obtained. Which can be used to find the same grid as that obtained in the encoder. (6) Effect corresponding to claim 8: Further, when detecting a moving contour, a contour on a reference image used for prediction is detected, and a reference image used for prediction and at least one frame other than that are already processed. The motion of the contour is estimated using the decoded image, and the contour with motion is selected by this,
The moving region can be accurately obtained, and the same grid as that obtained by the encoding device can be obtained using this. (7) Effects corresponding to claims 1 to 8: If a good prediction image is obtained, the amount of code generated in the subsequent encoding process of the prediction error is small, and the entire image is encoded with a small amount of data. be able to. Further, even when a sufficient data amount cannot be used as a code for the prediction error due to a small usable data capacity, a small transmission capacity, etc., a good predicted image results in a good decoded image. Therefore, according to the present invention, it is possible to provide a picture coding apparatus and a picture decoding apparatus that solve the problems of the conventional method and provide good image quality with less distortion.

[Brief description of drawings]

FIG. 1 is a configuration diagram for explaining an embodiment of an image encoding device according to the present invention.

FIG. 2 is a configuration diagram of a moving contour detection unit in FIG.

FIG. 3 is a configuration diagram of a moving area detection unit in FIG.

FIG. 4 is another configuration diagram of the moving contour detection unit in FIG.

FIG. 5 is a configuration diagram for explaining an example of an image decoding apparatus according to the present invention.

[Fig. 6] Fig. 6 is a diagram for describing motion compensation prediction by geometric transformation.

FIG. 7 is a configuration diagram of a predictive image generation circuit in a conventional image encoding device.

FIG. 8 is a diagram showing an example of a modified grid on a reference image.

FIG. 9 is a configuration diagram of a prediction image creation circuit in a conventional image decoding device.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Moving contour detecting unit, 2 ... Lattice point determining unit, 3 ... Motion estimating unit, 4 ... Predicting unit, 5 and moving region detecting unit, 6 ... Contour detecting unit,
7 ... Moving contour selecting unit, 8 ... Absolute value calculating unit, 9 ... Region detecting unit, 10 ... Motion estimating unit, 11 ... Moving contour detecting unit, 12 ... Lattice point determining unit, 13 ... Motion vector decoding unit, 14 ... Prediction Department,
15 ... Moving contour selection section.

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] September 2, 1994

[Procedure Amendment 1]

[Document name to be corrected] Drawing

[Correction target item name] Figure 8

[Correction method] Change

[Correction content]

[Figure 8]

Claims (8)

[Claims] 1. A predictive image creating circuit in a moving picture predictive coding apparatus, wherein an image is divided into rectangular blocks by a rectangular grid, and the shape of the rectangular block is deformed using motion vectors of the grid points to obtain a predictive image. A moving contour detecting means for detecting a moving contour by using a reference image used for prediction and a decoded image of at least one frame other than the reference image, and a contour matching the contour obtained by the moving contour detecting means. A grid point determining means for setting grid points on the reference image, and a motion estimating means for estimating at which position on the target image the grid point on the reference image obtained by the grid point determining means has moved, An image coding apparatus, comprising: a prediction unit that obtains a predicted image using the grid points on the reference image, a motion vector, and the reference image. 2. A moving area detecting means for detecting a moving area using a reference image used for prediction and a decoded image of at least one frame other than the reference image, the moving area detecting means for predicting. The image coding apparatus according to claim 1, wherein the image coding apparatus is configured by a contour detecting unit that detects a contour on a reference image to be used and a moving contour selecting unit that selects a moving contour. 3. The moving area detection means calculates a difference between a reference image used for prediction and a decoded image other than the reference image which has already been processed, and an absolute value of difference data obtained by the difference means. 3. The image coding apparatus according to claim 2, wherein the image coding apparatus comprises an absolute value calculating means and an area detecting means for detecting an area from the absolute value by the absolute value calculating means by threshold processing. 4. A motion estimation means for estimating the motion of the contour using the moving contour detection means, using a reference image used for prediction and a decoded image of at least one frame other than the reference image. And a moving contour selecting means for selecting a certain contour.
The image encoding device described. 5. An image is divided into quadrilateral blocks by a quadrilateral lattice, the quadrilateral block is deformed using the motion vector of the lattice points, and a prediction is performed in a prediction image generation circuit in a moving image prediction decoding apparatus that makes a prediction image. A moving contour detecting means for detecting a moving contour using a reference image and a decoded image of at least one frame other than the reference image, and a contour obtained by the moving contour detecting means. A grid point determining means for setting grid points on the reference image, a motion vector decoding means for extracting and decoding a motion vector incorporated in the encoded data, a grid point on the reference image, a motion vector and a reference An image decoding apparatus, comprising: a prediction unit that obtains a predicted image using an image. 6. A moving area detecting means for detecting a moving area using a reference image used for prediction and a decoded image of at least one frame other than the reference image, the moving area detecting means The image decoding apparatus according to claim 5, wherein the image decoding apparatus is configured by a contour detecting unit that detects a contour on a reference image to be used and a contour selecting unit that selects a moving contour. 7. The moving area detecting means calculates difference between a reference image used for prediction and a decoded image other than the reference image which has already been processed, and an absolute value of difference data obtained by the difference means. 7. The image decoding apparatus according to claim 6, wherein the image decoding apparatus comprises an absolute value calculating means and an area detecting means for detecting an area by threshold processing from an absolute value by the absolute value calculating means. 8. The moving contour detection means estimates a movement of the contour using a reference image used for prediction and a decoded image of at least one frame other than the reference image; The image decoding apparatus according to claim 5, further comprising moving contour selecting means for selecting.