JPH1023404A - Picture encoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To efficiently encode a picture in an optional shape by encoding a picture block in which a null area is buried by padding. SOLUTION: A padding means 12 identifies a picture area and a null area within the block with respect to inputted block data by shape information s14 in the picture block inputted from outside. Then a minimum rectangular area (pounding box) surrounding the picture area within the block is detected and segmented. The horizontal and vertical sizes of the pounding box are extended in optional directions to be powers of two. Then the processing for burying the null area (padding) by using data in the picture area is executed. Namely, padded pounding box is expanded horizontally and vertically into n×n (n is a power of two) by multiplying a power of two. A picture in an optional shape can be encoded by adding the padding means like this.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for encoding / decoding an image of an arbitrary shape, and relates to an efficient image encoding.

[0002]

2. Description of the Related Art Conventionally, there has been a technique described in the following literature as a technique in such a field. Reference: Hiroshi Yasuda, "International Standard for MPEG / Multimedia Coding", published by Maruzen Co., Ltd., September 30, 1994, pp.60-75. The document describes a method for encoding an image signal. FIG. 2 is a block diagram of an image encoding device described in the above-mentioned document. This conventional image coding apparatus includes a block dividing unit 21 for dividing an image signal into n × n blocks.
And encoding means 22 for performing compression encoding in block units.

An input image signal s20 is connected to a block dividing means 21, and an output s21 of the block dividing means is encoded by a coding means 22.
And the output s22 of the encoding means is output to the output terminal.

The block dividing means 21 shown in FIG. 2 divides an input two-dimensional image signal s20 into n × n blocks and outputs the divided signals.

[0005] The encoding means 22 performs, for example, two-dimensional block cosine transform on the input block data, performs quantization and variable length encoding, and outputs the code.

[0006]

However, the conventional image coding apparatus has the following problems. The size of the input image signal in the horizontal and vertical directions must be a positive integer multiple of the block size. Therefore, an image of an arbitrary shape cannot be encoded as it is. In order to perform encoding, it is necessary to surround an image of an arbitrary shape with a block and fill a blank area without image data in the resulting block image. Commonly conceived filling methods include, for example,
Methods such as filling the average value of the image area or copying and filling the pixels at the edges of the image area are conceivable.However, in these methods, a large gap is created between the image area and the blank area, Many high-frequency components are often generated, which greatly affects the coding efficiency.

[0007]

A block dividing unit for dividing an input image signal into blocks to generate an image block, and a blank in the image block based on shape information of the image signal input from outside. The image processing apparatus includes a padding unit that fills an area, and an encoding unit that encodes the image block in which a blank region is filled by the padding unit.

[0008]

BEST MODE FOR CARRYING OUT THE INVENTION

<< Specific Example 1 >><Structure> Hereinafter, a specific example 1 according to the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram of an image coding apparatus according to a first embodiment of the present invention. The image coding apparatus according to the first specific example of the present invention is different from the conventional image coding apparatus in that a block is provided between the block dividing means 11 and the coding means 13 on the basis of externally provided image shape information. And padding means 12 for filling a blank area where no image data exists.

In FIG. 1, an input image signal s10 is connected to a block dividing means 11, and an output s11 of the block dividing means is provided.
Further, the shape information s14 of the image input from the outside is connected to the padding means 12, the output s12 of the padding means is connected to the coding means 13, and the output s13 of the coding means is outputted to the external terminal.

<Operation> Since the block dividing means 11 and the encoding means 13 have the same functions as those of the prior art and operate in the same manner, the description will be omitted. Hereinafter, the operation of the padding means 12 will be described with reference to FIG. The padding means 12 identifies an image area and a blank area in the input block data, as shown in FIG. 3A, based on the externally input shape information s14 of the image block. Then, as shown in FIG. 3B, the smallest rectangular area surrounding the image area in the block (hereinafter, this area is called a bounding box) is detected and cut out. Then, as shown in FIG. 3C, the bounding box is expanded in an arbitrary direction (here, the lower right) so that the horizontal and vertical sizes become powers of two. Then, as shown in FIG. 3D, a process of filling a blank area using data of an image area (hereinafter, referred to as padding) is performed. Then, as shown in FIG. 3 (e), the padded bounding box is expanded by a power of 2 in the horizontal and vertical directions so that it becomes n × n (n is a power of 2). As the padding process, for example, the following method is used.

1. First, in the horizontal direction, if all lines are blank or all image areas, the process proceeds to the next line without processing. If a blank area in the row is sandwiched between image areas, the blank area is filled using the average of the image data on both sides or a weighted average corresponding to the distance. If there is a blank area on that line on either side of the bounding box, the image data on the opposite side of the blank area is copied to fill the blank area. This process is performed for all rows.

2. Then, the same processing as in the horizontal direction is performed in the vertical direction to fill all the blank areas. Also,
The padding process can be performed as follows.

1. First, the above 1 is independently set in the horizontal and vertical directions.
And the average of both image data is taken.
In this case, if both are not blank areas, the average is output. If either is a blank area,
Outputs data for non-blank areas. If both are blank areas, output as blank areas.

2. Then, the process of step 1 is repeated again to fill all the blank areas.

[0015] As the processing for expanding the bounding box, first, if the block size is n = 2m and the horizontal size of the bounding box is k = 2s, the horizontal direction of the bounding box is expanded by 2m-s times. .
Similarly, if the vertical size of the bounding box is l = 2t in the vertical direction, the vertical direction of the bounding box is expanded by 2m-t times. For example, this 2x expansion processing can be realized by repeating the interpolation processing for interposing the average of two adjacent pixels x times.

<Effect> As described above, the first aspect of the present invention is as follows.
According to the specific example, by adding padding means to the conventional image coding apparatus, it becomes possible to code an image of an arbitrary shape. Also, in comparison with the simple pixel expansion and the processing of filling the average pixel value, in the first specific example of the present invention, the use of the bounding box detecting means, the bounding box expanding means and the area expanding means makes it possible to always use the inside of the block. Can be padded so that the correlation between the pixels becomes higher, so that the coding efficiency of the block can be improved.

<< Embodiment 2 >><Configuration> Hereinafter, Embodiment 2 according to the present invention will be described in detail with reference to the drawings. The image coding apparatus according to the second embodiment of the present invention is characterized in that, in the padding means of the image coding apparatus according to the first embodiment of the present invention, the area expansion means by pixel copy or pixel interpolation shown in FIG. This is an area expanding means by wrapping copy. FIG. 4 is an explanatory diagram of a padding method of the image encoding device according to the second specific example of the present invention.

<Operation> In the coding apparatus according to the second embodiment of the present invention shown in FIG. 4, (a) to (d) perform the same operation as the first embodiment of the present invention, Then, as shown in FIG. 4E, the padding bounding box is subjected to 2m-s times and 2m-t times repeated copying in the horizontal and vertical directions to extend the area.

<Effects> As described above, according to the second embodiment of the present invention, in the first embodiment of the present invention,
By making the area expansion method a wrapping copy method,
It is possible to further enhance the coding efficiency of a coding unit having an even symmetric property such as DCT or DFT (Discrete Fourier Transform).

<< Embodiment 3 >><Structure> The image coding apparatus according to the third embodiment of the present invention is different from the padding means of the image coding apparatus according to the first embodiment of the present invention in that the padding means shown in FIG. The area expanding means by pixel interpolation is changed to the area expanding means by pixel copying.

<Operation> In the padding means of the third embodiment of the present invention, the same operations as those of the first embodiment of the present invention are performed from (a) to (d) of FIG. For a padded bounding box, one pixel is copied horizontally and vertically 2ms-times and 2mt-times, respectively, to extend the area.

<Effects> As described above, according to the third embodiment of the present invention, in the first embodiment of the present invention,
By using the pixel copy method for the area extension method, it is possible to further enhance the coding efficiency of the coding means using a conversion means capable of efficiently expressing the local characteristics of an edge such as a wavelet transform.

<< Embodiment 4 >><Structure> The image coding apparatus according to the fourth embodiment of the present invention is characterized in that in the padding means of the image coding apparatus according to the first, second and third embodiments of the present invention, The method of expanding the bounding box is changed to an adaptive expansion method that changes the expansion direction according to the image area.

<Operation> The method of expanding the bounding box in the padding means will be described below. Other means operate in the same manner as the first or second embodiment of the present invention.

In the bounding box expansion method according to the fourth embodiment of the present invention, first, the ratio of the image area to the leftmost and rightmost columns in the bounding box is calculated, and the ratio of the image area is large. Extend the bounding box horizontally. Similarly, in the vertical direction, the ratio of the image area to the upper and lower rows in the bounding box is calculated, and the vertical bounding box is expanded in the direction in which the image region occupies a larger percentage. In general, when the end of the image area changes more drastically than the inside of the image area and is extended using the image data at the end, many high-frequency components are generated, and the coding efficiency is reduced. On the other hand, at the edge of the bounding box, the higher the proportion occupied by the image area is, the higher the probability of being inside the image area, and the higher the probability that the inside of the image is smooth, so by expanding the area in that direction, A smoother image is obtained, and thus higher encoding efficiency is obtained in the subsequent encoding means.

<Effects> As described above, according to the fourth embodiment of the present invention, in the first, second and third embodiments of the present invention, the method of expanding the bounding box in the padding means is used. By expanding the image in a direction in which the proportion of the image area occupies a larger portion, the image in the bounding box can be made smoother, and thus the coding efficiency of the block can be improved.

[0027]

According to the present invention, a block dividing means for dividing an input image signal into blocks to generate an image block, and a blank in an image block based on shape information of an externally input image signal. With a configuration including padding means for filling the area and coding means for coding the image block in which the blank area is filled by the padding means, it is possible to code an image of an arbitrary shape. Also, compared to simple pixel expansion or processing such as filling in the average pixel value, by using the bounding box detection means, the bounding box expansion means, and the area expansion means, the correlation between the pixels in the block is always increased. Since padding can be performed, the coding efficiency of the block can be increased.

Further, the area is extended by folding and copying the bounding box with respect to any one of the four sides of the bounding box, so that the area extending means of the present invention is set to the folding copy method, and DCT, DFT, etc. It is possible to further improve the coding efficiency of the coding means having the even symmetric characteristic. Further, by expanding the area by copying pixels in the bounding box in the horizontal or vertical direction, the area expansion means of the present invention is set to a pixel copy method, and a conversion that can efficiently express local characteristics of edges such as wavelet conversion. The encoding efficiency of the encoding means using the means can be further improved. Furthermore, the ratio of the image area at both ends of the bounding box is determined, and by expanding in the direction in which the image area ratio is large, the bounding box extending means in the padding means is expanded to the direction in which the image area occupies a large amount. The image in the bounding box can be made smoother, and thus the coding efficiency of the block can be improved.

[Brief description of the drawings]

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

FIG. 2 is a block diagram of a conventional image encoding device.

FIG. 3 is an explanatory diagram of a padding method according to a first embodiment of the present invention.

FIG. 4 is an explanatory diagram of a padding method according to a second embodiment of the present invention.

[Explanation of symbols]

 11 block dividing means 12 padding means 13 encoding means

Claims (6)

[Claims] 1. A block dividing means for dividing an input image signal into blocks to generate an image block, and padding for filling a blank area in the image block based on shape information of the image signal input from outside. And an encoding unit for encoding the image block in which a blank area is filled by the padding unit. 2. The image processing apparatus according to claim 1, wherein the padding unit includes: an identification unit that identifies an image area in the image block;
Bounding box detecting means for detecting and cutting out the smallest rectangular bounding box surrounding the image area, bounding box expanding means for expanding the bounding box, and bounding box padding means for filling a blank area in the expanded bounding box. 2. The image encoding apparatus according to claim 1, further comprising: an area expansion unit that expands a bounding box in which a blank area in the bounding box is filled to the size of the image block. 3. The image coding apparatus according to claim 1, wherein the area expansion unit expands an area by copying pixels in the bounding box in a horizontal or vertical direction. 4. The apparatus according to claim 1, wherein the area expansion unit expands the area by inserting pixels between adjacent pixels in the bounding box and substituting an average value of the adjacent pixel data. The image encoding device according to claim 1. 5. The area expansion unit according to claim 1, wherein the area expansion unit expands the area by folding and copying the bounding box on any one of four sides of the bounding box. Image coding device. 6. The image processing apparatus according to claim 1, wherein the bounding box expanding unit obtains a ratio of an image area at both ends of the bounding box, and expands the image area in a direction in which the image area ratio is large. Image coding device.