JPH0879756A - Moving image encoding device - Google Patents

Abstract

PURPOSE: To reduce the code amount of data without degrading visual picture quality by matching a moving 'area' and the 'area' which is a unit for changing the accuracy of encoding and visually reducing distortion and to the discontinuity of encoding accuracy. CONSTITUTION: An area division part 10 divides image data into the plural areas corresponding to the direction and size of a motion vector. An encoding part 20 performs the encoding for the respective areas, performs the encoding with high compressibility (low accuracy) when the motion vector of the area is large and performs the encoding with low compressibility (high accuracy) when the motion vector of the area is small inversely. Thus, this moving image encoding device capable of reducing the data amount of codes without degrading the visual picture quality so much is obtained.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture coding device. In particular, the present invention relates to a moving picture coding apparatus capable of realizing a high compression rate by adapting to each scene.

[0002]

2. Description of the Related Art In recent years, with the progress of image processing technology, not only still images but also moving images have been extensively processed. For example, various kinds of image processing are performed in image communication such as videophone and videoconference, or processing of video signals.

Since the data amount of such a moving image is generally extremely large, various moving image coding methods for compressing the data amount of the moving image have been proposed. Coding for compression of a moving image is one of general methods, for example, for data compression by utilizing the correlation of image signals in space and time. There is also proposed a method of compressing data by utilizing a part of human visual characteristics.

The former method of utilizing the correlation of data, that is, the method of utilizing statistical properties, utilizes the fact that the data of a moving image has a very strong correlation between frames or between pixels. For example, as this type of encoding method, a predictive encoding method can be cited. The predictive coding system predicts a pixel value of interest from a previous frame or surrounding pixels and codes a prediction error. It is expected that this prediction error will be a scattered value centered on zero. As another encoding method of this kind, a conversion encoding method can be cited. This transform coding method is a coding method that utilizes the fact that when moving image data is converted into the spatial frequency domain, the energy concentrates in the low frequency domain. That is,
This is a method in which an important component in the low frequency region and an unimportant component in the high frequency region are separated to provide a difference in encoding accuracy.

An example of the latter method of utilizing the human visual characteristic is a method of utilizing the fact that the human visual sense is insensitive to high frequency noise. this is,
This is a method of roughly (with low accuracy) encoding the high-frequency component of moving image data. In addition, since human vision is generally sensitive to a luminance component but insensitive to a color difference component, a method of roughly coding (with low accuracy) a color difference signal may be considered.

Other usable human visual characteristics include a visual characteristic that is sensitive to noise in a flat portion but insensitive to noise in an edge portion, and a characteristic that sensitivity to an oblique component is low. Conventionally known.

[0007] In particular, a human visual characteristic peculiar to a moving image includes a perceptual characteristic of motion. That is, human vision is insensitive to objects moving at a certain speed or higher. Further, as the speed of the object increases, the visual sensitivity to high frequency components deteriorates. Further, the moving body visual acuity deteriorates as the illuminance decreases. For example,
When the eyes are fixed, the change in viewing angle per second is 15 deg.
It is generally known that the shape of the object is perceived for movements up to / sec, but it appears as a mere band of light for movements over 50 deg / sec. It is known that when the eyes are moved, it is possible to follow the movement up to about 25 to 30 deg / sec. For example, such visual characteristics could be used in sports broadcasting and in the coding of images of car races.

There is also a visual characteristic due to human psychological factors. For example, human vision is considered to be insensitive to the distortion of a portion (background) other than the attention portion.

A moving picture coding apparatus utilizing the above human visual characteristics is described in, for example, Japanese Patent Laid-Open No. 3-16490. The moving picture coding device described here detects a motion vector between frames of a moving picture signal and determines whether or not the block is moving based on this motion vector, and determines that the block is moving. In this case, the coding device sets a coarse coding characteristic, and sets a dense coding characteristic when it is determined that the block is not moving. That is, the accuracy of encoding is changed depending on whether or not each block is moving to compress data.

[0010]

DISCLOSURE OF THE INVENTION Problems to be Solved by the Invention
According to the moving picture coding apparatus described in Japanese Patent Laid-Open No. 90, since the coding can be performed in consideration of the motion of the picture, the data can be efficiently compressed without deteriorating the visual quality of the reproduced moving picture. Is.

However, in the above-described conventional moving picture coding apparatus, it is judged for each block whether or not it is moving. Therefore, even in a block which is determined to be moving, there may be a static part in it. In such a case, the image quality of the still portion may be visually deteriorated. On the contrary, even in a block determined to be still, there may be a moving part. In such a case, the moving part may be encoded with high accuracy and the data amount may increase.

Further, since the boundary between the actual moving / static region and the boundary between the blocks do not match, discontinuity occurs in the coding accuracy at the boundary between the moving / static regions, resulting in distortion.

The present invention has been made in view of the above problems, and an object thereof is to match a moving "area" with an "area", which is a unit for changing the coding accuracy, and to make the coding accuracy. It is an object of the present invention to provide a moving picture coding apparatus capable of visually reducing the distortion due to the discontinuity and reducing the coding amount of data without deteriorating the visual image quality.

[0014]

In order to solve the above-mentioned problems, a first aspect of the present invention is to divide an area of a moving image into a plurality of irregular areas, and a motion vector of each of the divided areas. And a coding unit that performs coding for each of the areas with accuracy according to the size of the moving picture coding apparatus.

In order to solve the above-mentioned problems, a second aspect of the present invention provides an area dividing means for area-dividing a moving image into a plurality of irregular areas, and an accuracy according to the brightness of each of the divided areas.
And a coding unit that performs coding in each of the areas.

In order to solve the above-mentioned problems, a third aspect of the present invention is the moving image encoding apparatus according to the first or second aspect of the present invention, wherein the area dividing means is a direction of movement of each pixel in the image. A moving picture coding apparatus characterized by dividing a plurality of areas according to size.

In order to solve the above-mentioned problems, a fourth aspect of the present invention is the moving image encoding apparatus according to the first or second aspect of the present invention, wherein the area dividing means has a magnitude of luminance of each pixel in the image. It is a moving picture coding device characterized by dividing a plurality of regions according to the size.

A fifth aspect of the present invention, in order to solve the above-mentioned problems, in the moving image encoding apparatus according to the first or second aspect of the present invention, the area dividing means is arranged according to the color of each pixel in the image. The moving picture coding apparatus is characterized in that a plurality of areas are divided into two areas.

[0019]

The area dividing means in the first aspect of the present invention divides an image into irregular areas, and the encoding means encodes the accuracy, that is, the compression rate, according to the magnitude of the motion vector of the area. Therefore, the discontinuity of the encoding accuracy along the block boundary does not occur as compared with the case where the block is divided into rectangular block areas and encoded.

The area dividing means in the second aspect of the present invention divides the image into irregular areas, and the encoding means encodes with accuracy according to the magnitude of the luminance of the area, that is, with a compression rate. Therefore, as compared with the case of dividing into a rectangular block area and encoding, discontinuity of the encoding accuracy along the boundary of the block does not occur, and since the encoding accuracy is changed by the brightness, The amount of encoded data can be reduced by lowering the encoding accuracy for a region having extremely low brightness.

The area dividing means in the third aspect of the present invention divides the area according to the direction and size of the movement of each part of the image. That is, a portion that moves in the same (similar) direction is grasped as one area, and the encoding means performs a predetermined encoding on the thus grasped area.

The area dividing means in the fourth aspect of the present invention divides the area according to the brightness of each pixel of the image. That is,
A plurality of pixels having the same (similar) brightness are grasped as one region, and the encoding means performs a predetermined encoding on the region thus grasped.

The area dividing means in the fifth aspect of the present invention divides the area according to the color of each pixel of the image. That is, a plurality of pixels having the same (similar) color are grasped as one region, and the encoding means performs a predetermined encoding on the region thus grasped. Here, color means chromaticity,
A value such as saturation and color difference.

[0024]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a moving picture coding apparatus which is a preferred embodiment of the present invention. FIG.
As shown in, the moving image data is first divided into a plurality of predetermined regions by the region dividing unit 10. The division of this area is performed by luminance, color, motion vector (for example, optical flow), and the like. Here, the color specifically refers to values such as chromaticity, saturation, and color difference. What is characteristic of the present embodiment is that this area division is performed in an indeterminate area, and is not simply division into rectangular blocks. As described above, in the conventional encoding device that simply divides into rectangular blocks, block-shaped distortion is likely to occur, but in the present embodiment, the predetermined feature amount (luminance, color, optical flow) is used. And the like), the block-shaped distortion as in the conventional case does not occur because it is divided into amorphous areas.

When area division is performed by the brightness of pixels,
Region division is performed by clustering into pixel groups having similar luminance values, but as a concrete method, for example, a K-means algorithm or the like is suitable. K
The averaging algorithm is a method of area division by clustering, and is a method of classifying a set of pixel groups having certain R-type feature values X1 to XR into K clusters. The outline of the operation of this K-means algorithm will be briefly described below.

(Step 1) K initial cluster center values Z1 to ZK are determined. This is one of the R types of feature values.
For example, arbitrary K pieces are selected as initial values.

(Step 2) R type feature value set {X
1, ~, XR} are converted into K classes Sk (k = 1 to 1) as follows.
K).

Now, the characteristic value Xr (r = 1 to R) is k
For (= 1 to K) | Xr-Zk | = min {| Xr-Zm |} (m = 1,
2, the characteristic value Xr is used as an element of the class Sk. That is, each feature value Xr
Is the central value Zk (k = 1 to 1) whose feature value Xr is the closest.
It is classified into the category Sk of K).

(Step 3) The central value Zk (k = 1 to K) of each of the divided Sks is newly updated as follows.

New Zk = (1 / nk) ΣXr (sum of all Xr elements of Sk) where nk is the number of feature values Xr belonging to the class Sk. The new center value Zk thus obtained is the class Sk
Is the average value of the feature values Xr belonging to
It represents Xr belonging to the class Sk.

(Step 4) Z in Step 3
Steps 2 to 3 are repeated until the value of Zk does not change even when the value of k is updated.

The K-means algorithm is executed as described above. If a brightness value is adopted as the characteristic value, the area is divided by the brightness value, and if the color is a characteristic value, the area is divided by color. If saturation is used as this color, division by saturation is performed, and if chromaticity is used, division by chromaticity is performed. Needless to say, if the color difference is used, the division is performed according to the color difference. The same applies to the optical flow.

There is another simple method for area division, for example, an area expansion method. In this method, if there are adjacent areas having similar attribute values, they are integrated to form a larger area.

After the area is divided into irregular areas in this way, encoding is performed for each area. This encoding is shown in FIG.
Is performed by the encoding unit 20. This encoding unit 20
Is performed by, for example, DCT encoding.

A feature of this embodiment is that the encoding unit 20 performs encoding with a different compression rate, that is, accuracy for each area. That is, the visually important region is compressed (encoded) with high accuracy, and the non-important region is encoded with low accuracy, that is, a high compression rate. In this way, since the encoding accuracy is changed according to the importance of each area, efficient encoding can be performed,
It is possible to reduce the amount of encoded data. As a result, the amount of data to be transmitted can be reduced in a videophone or the like, and a moving image that can sufficiently follow the movement of an object can be provided.

In the present embodiment, the setting of the coding accuracy for each area is performed on the basis of, for example, the magnitude of the motion vector and the brightness of each area.

For example, when the motion vector is used as a reference, the accuracy is low for an area having a large motion vector.
That is, encoding is performed with a high compression rate. this is,
This is because the human visual sensitivity to moving objects is generally reduced. In other words, even if the quality of a moving image deteriorates to some extent, it is often impossible to identify it with the naked eye. FIG. 2 shows a schematic diagram in the case where a moving area and a non-moving area exist in one screen in this way. As shown in FIG. 2, there is a region 30 having no motion as a background in one image, while a region 40 having a certain motion like an automobile also exists.

As described above, in this embodiment, each region is encoded at a high compression rate (with low accuracy) for a region having a large motion vector size,
If the region where the motion vector is small is encoded at a low compression rate (high accuracy), it is possible to reduce the amount of code data without significantly deteriorating the image quality of human eyes.

The detection of the motion vector of each area is performed by the inter-frame matching for each area or the difference of each area between frames.

It is also preferable to change the compression rate of coding according to the brightness of each area. According to the present embodiment, a region with low luminance is encoded with a high compression rate (with low precision), and a region with high luminance is encoded with a low compression rate (with high precision). This is because the human eyes cannot see the dark part well, so even if the image quality of the dark part deteriorates, it has less effect on the "visual appearance" of the naked eye than the deterioration of the image quality in the high brightness region. is there. In the present embodiment, since the image compression rate is changed according to the brightness of each area in this way, it is possible to reduce the data amount of the code without causing much deterioration of the image quality by the human eyes. is there.

For setting the compression rate, it is also preferable to use, for example, the size of optical flow.
It has substantially the same action and effect as the above-described motion vector and the like.

As described above, according to the present embodiment, the moving image is divided into the irregular areas, and the encoding is performed with a different compression rate (accuracy) for each area. It is possible to obtain a moving image coding apparatus capable of reducing the data amount of codes without causing the above-mentioned problem and without causing the deterioration of the visual quality of human eyes.

[0044]

As described above, according to the first aspect of the present invention, the coding can be performed without causing the distortion due to the block shape, and the efficient coding in consideration of the motion vector of the image area can be performed. The effect is obtained that a moving picture coding apparatus capable of

As a result, it is possible to obtain a videophone or the like in which the code data amount is reduced and the image quality is improved without degrading the image quality with the naked eye.

According to the second aspect of the present invention, the coding can be performed without causing the distortion due to the block shape, and
The present invention has the effect of providing a moving picture coding apparatus that enables efficient coding in consideration of image brightness.

As a result, it is possible to obtain a videophone or the like in which the code data amount is reduced and the image quality is improved without degrading the image quality with the naked eye.

According to the third aspect of the present invention, since the area division is performed so that the portion having the similar motion is defined as one area, it is possible to perform the area division in conformity with the contour of the object existing in the image, By changing the coding characteristics for each area,
An effect is obtained that a moving picture coding apparatus capable of more efficient coding is obtained.

According to the fourth aspect of the present invention, since the area division is performed so that the portion having the same brightness is regarded as one area, it is possible to perform the area division in conformity with the contour of the object existing in the image. By changing the coding characteristics for each area,
An effect is obtained that a moving picture coding apparatus capable of more efficient coding is obtained.

According to the fifth aspect of the present invention, since the area division is performed so that the portions having the same color are regarded as one area, it is possible to perform the area division in conformity with the contour of the object existing in the image. By changing the coding characteristic for each area, it is possible to obtain a moving image coding apparatus capable of more efficient coding.

Further, by combining the first and third aspects of the present invention, the coding corresponding to the movement of the area is performed in each area, so that the visual quality of the image is not deteriorated and the efficiency is improved. An effect is obtained that a moving picture coding apparatus capable of general coding is obtained.

Further, by combining the second and fourth embodiments, the coding corresponding to the brightness of the area is performed in each area, so that the efficiency is improved without causing the visual quality deterioration. An effect is obtained that a moving picture coding apparatus capable of general coding is obtained.

[Brief description of drawings]

FIG. 1 is a conceptual diagram of functional blocks of a moving picture coding apparatus that is a preferred embodiment of the present invention.

FIG. 2 is a schematic diagram in the case where two regions, a region having a motion and a region having no motion, exist in one image.

[Explanation of symbols]

 10 area division section 20 encoding section

Claims (5)

[Claims] 1. A region dividing means for dividing a moving image into a plurality of irregular regions, and a code for encoding each region with accuracy according to the magnitude of the motion vector of each divided region. A moving picture coding device, comprising: 2. A region dividing unit that divides a moving image into a plurality of irregular regions, and an encoding unit that encodes each region with accuracy according to the brightness of each divided region. A moving picture coding device comprising: 3. The moving picture coding apparatus according to claim 1, wherein the area dividing means divides a plurality of areas in accordance with the direction and size of the movement of each pixel in the image. A featured moving image encoding device. 4. The moving picture coding apparatus according to claim 1, wherein the area dividing means divides a plurality of areas according to the brightness of each pixel in the image. Video encoding device. 5. The moving picture coding apparatus according to claim 1, wherein the area dividing unit divides a plurality of areas according to a color of each pixel in the image. Encoding device.