JP2002271793A - Image compression encoder and method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce deterioration in the image quality at a border of an area designated with high image quality in a moving picture. SOLUTION: A discrete wavelet converter 12 applies two-dimensional discrete wavelet conversion to an input image signal. An area designation device 14 outputs mask information denoting an area whose image quality is to be increased. An area surrounding calculation device 16 calculates a surrounding part of a border within a designated area. A low-order bit reduction device 18 replaces low-order N bits of conversion coefficient data of the surrounding part outputted from the converter 12 with zero. A quantization circuit 20 quantizes the conversion coefficient data and shifts up the data in the designated area in the obtained quantization index by a specified number of bits. An entropy encoder 22 decomposes the quantization index received from the quantization circuit 20 into bit planes and applies binary arithmetic coding to the bit plane in the unit of bit plane to provide an output of a code stream.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image compression encoding apparatus and method for compressing and encoding an input image.

[0002]

2. Description of the Related Art In recent video movies, a digital recording system using discrete cosine transform has become mainstream. That is, the image captured by the solid-state imaging device is decomposed into blocks, subjected to discrete cosine transform, the obtained coefficients are quantized, rearranged in order of frequency and rearranged from the low-frequency range to variable-length coding, and coefficients at distant positions on the screen. Are packed together in a fixed-length sync block to perform compression with no noticeable deterioration, and are recorded on a magnetic tape at a fixed rate.

As a next-generation image compression method, application of a compression method using a discrete wavelet transform to moving images is being studied. In JPEG2000, etc., a high importance part is designated as an area, and the compression rate of that part is set lower than other parts, thereby achieving a high compression rate as a whole and achieving high image quality of a desired part. Alternatively, recording can be performed at a low compression ratio including non-compression.

[0004]

However, when a high-quality area (or an uncompressed area) is specified in a moving image, there are cases where the subject enters and does not enter the area depending on the movement of the subject in that area. The image quality of the background changes drastically, and the boundaries become noticeable. This gives the viewer discomfort, and consequently,
The visual quality is greatly degraded.

[0005] It is an object of the present invention to provide an image compression encoding apparatus and method for solving such a disadvantage.

[0006]

According to the present invention, there is provided an image compression encoding apparatus, comprising: an area setting means for setting a predetermined area of an input image signal; and a predetermined area and another area set by the area setting means. And a boundary portion detecting means for setting a boundary region corresponding to the boundary portion, compression encoding the image signal of the predetermined region at a first compression ratio, and converting the image signal of the boundary region to Coding means for compression-coding at a second compression rate higher than the first compression rate, and compression-coding the image signal of the other area at a third compression rate higher than the second compression rate; It is characterized by having.

According to the image compression / encoding method of the present invention, an area setting step of setting a predetermined area of an input image signal, and a boundary between the predetermined area and another area set in the area setting step are performed. Detecting and setting a boundary area corresponding to the boundary area; and compressing and encoding the image signal of the predetermined area at a first compression rate, and converting the image signal of the boundary area to the first compression rate. Compression coding at a higher second compression rate, and compression coding the image signal of the other area at a third compression rate higher than the second compression rate. And

[0008]

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

FIG. 1 is a schematic block diagram showing an embodiment of the present invention. 10 is an image input device, 12 is a discrete wavelet transform device, 14 is a region designation device, 16 is a region periphery calculation device, 18 is a lower bit deletion device, 20 is a quantization device, 22 is an entropy coding device, and 24 is a code output. Device.

[0010] Pixel signals constituting an image to be encoded are input to the image input device 10 in a raster scan order. The output of the image input device 10 is input to the discrete wavelet transform device 12. The discrete wavelet transform device 12 performs a two-dimensional discrete wavelet transform process on the input image signal, and outputs a transform coefficient to the lower bit removing device 18.

FIG. 2 shows a discrete wavelet converter 12.
FIG. 2 shows a schematic block diagram of the configuration. In the two-dimensional discrete wavelet transform, one-dimensional conversion is sequentially performed in the horizontal and vertical directions of an image. That is, the input image signal is subjected to wavelet transform processing in the horizontal direction, and is decomposed into low-pass coefficients and high-pass coefficients. Then, the data is thinned by half by downsizing (downward arrow). The above processing is repeated for components obtained by performing low-pass filter processing in the horizontal and vertical directions on the data of the thinning result.
As a result, coefficient data in which the amount of data in the low frequency region is reduced by frequency division in the horizontal and vertical directions is accumulated as the generated coefficient component.

FIG. 3 shows an example of a two-level transform coefficient group obtained by a two-dimensional wavelet transform process. HL for high frequency in the horizontal direction and low frequency in the vertical direction by the first division
1. LH1 is a low band in the horizontal direction and a high band in the vertical direction, and HH1 is a high band in the horizontal direction and a high band in the vertical direction. In the same manner as described above, the low frequency in the horizontal direction and the low frequency in the vertical direction are divided in the same manner as described above to obtain HL2, LH2, and HH2, and the low frequency in the horizontal direction and the low frequency in the vertical direction are designated as LL. In this way, the input image is a coefficient sequence HH of a different frequency band.
1, HL1, LH1, HH2, HL2, LH2, LL. In the following description, these coefficient sequences are called subbands.

An area specifying device 14 is an area (ROI: ReI) to be coded with high image quality in an image to be coded.
gion of interest),
When discrete wavelet transform is performed on the target image, mask information indicating which coefficient belongs to the specified area is generated.

FIG. 4 shows an example of generating mask information. FIG.
It is assumed that a star-shaped area is designated as shown in FIG. When the image to be encoded is subjected to the discrete wavelet transform, the region specifying device 14 calculates a region occupied by each of the specified regions in each subband, and sets the calculated region as mask information. FIG. 4 (b)
An example of the mask information calculated in this way is schematically shown. In the example shown in FIG. 4, the mask value is 1 in the star-shaped designated area, and the mask value is 0 in other areas. The entire mask information is the same as the configuration of the transform coefficients by the two-dimensional discrete wavelet transform. By taking the logical product of the mask information and the two-dimensional discrete wavelet transform coefficient of the image to be coded, the designated area portion can be extracted.

The mask information generated by the area designation device 14 is supplied to the area periphery calculation device 16 and the quantization device 20. The area specifying device 14 further calculates a bit shift amount B according to the image quality required for the specified area, and supplies the calculated bit shift amount B to the quantization device 20 together with the mask information.

The region periphery calculation device 16 calculates a portion near the boundary of the designated region, and outputs the calculation result to the lower bit deletion device 18. With reference to FIG. 5 and FIG. 6, a portion near the boundary calculated by the ellipse area periphery calculation device 16 will be described.
In FIG. 5, an elliptical region 30 surrounding a person is represented by JPEG20.
It is assumed that 00 is designated as an important area called ROI. In this case, the area surrounding calculation device 16 determines the designated area 3
A skin portion that is within 0 and is its outer periphery (periphery around the region)
32 is calculated. FIG. 6 is an image in which the person of interest has moved. In FIG. 6, the ROI region 30 and the region peripheral portion 3 are shown.
2 also moves with the movement of the person. As a result of the movement of the person, the background tree 34 gradually enters the ROI area 30,
Then, it exits from the ROI area 30. In the conventional example, the image quality greatly varies depending on whether or not the tree 34 is in the ROI area 30, but in the present embodiment, the area peripheral portion 32 is set in the ROI area 30, and as described in detail below. The processing makes the change in the image quality of the tree 34 inconspicuous.

Based on the calculation result of the region periphery calculation device 16, the lower bit removal device 18 removes the lower bits of the discrete wavelet transform coefficients belonging to the region periphery portion 32 in the ROI region 30 by a predetermined number N of bits. The result is supplied to the quantization device 20.

FIG. 7 shows an example in which the lower bit of the discrete wavelet transform coefficient is replaced with 0 by the lower bit removing unit 18. By deleting bits in advance by the lower bit deletion device 18, an intermediate image quality portion can be set between a portion having a higher degree of importance inside the region peripheral portion 32 and the outside of the ROI region 30. . Area peripheral portion 32
By removing the bits, more information is allocated to the other parts, and the overall image quality is improved.

First, the quantizer 20 quantizes the discrete wavelet transform coefficients supplied from the lower bit reducing unit 18 by a predetermined quantization step, and outputs an index for the quantized value. Quantizer 20 then:
The quantization index is changed based on the mask information and the shift amount input from the area specifying device 14. As a result, only the quantization index belonging to the spatial area specified by the area specifying device 14 is shifted up to the specified bit number information.

FIGS. 8 and 9 show examples of changes in the quantization index due to shift-up. In FIG. 8, three quantization indexes exist in each of three subbands. FIG. 9 shows a case where the mask value in the hatched quantization index in FIG. 8 is 1 and the shift number is 2,
This shows the quantization index after the shift.

The quantization index thus changed is applied to the entropy coding device 22. The entropy coding device 22 decomposes the quantization index input from the quantization device 20 into bit planes, performs binary arithmetic coding on a bit plane basis, and outputs a code stream.

The operation of the entropy coding device 22 will be described with reference to FIG. FIG. 10A shows an input of the entropy encoding device 22, and FIG. 10B shows an output.
In FIG. 10A, three non-zero quantization indices are present in a subband having a size of 4 × 4, and have values of +13, −6, and +3, respectively. . The entropy coding device 22 scans this area to find the maximum value M, and calculates the number of bits S required to represent the maximum quantization index. In the example shown in FIG. 10A, since the maximum coefficient value is 13, S is 4, and the 16 quantization indexes in the sequence are:
As shown in FIG. 10B, the processing is performed in units of four pit planes.

The most entropy coder 22 first
Binary arithmetic coding of each bit of the most significant bit plane,
The data is output to the code output device 24 as a bit stream.
Next, the bit plane is lowered by one level, and similarly, each bit in the bit plane is encoded until the target bit plane reaches the least significant bit plane.
4 is output. At this time, when the first non-zero bit is detected in the bit plane scanning, the code of the quantization index is entropy-coded immediately after that.

Although a monochrome image has been described as an example, a color image of an RGB signal is encoded independently for each color component.

In the above embodiment, the boundary part inscribed in the designated area is set. However, it is needless to say that the same applies when the boundary part circumscribed in the designated area is set.

[0026]

As can be easily understood from the above description, according to the present invention, the boundary of the designated area can be set to an intermediate image quality, so that the change in the image quality of the image part entering and exiting the designated area can be reduced. it can. As a result, the visual image quality can be improved and the overall image quality can be improved.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram of an embodiment of the present invention.

FIG. 2 is a schematic configuration block diagram of a discrete wavelet transform device 12;

FIG. 3 is an explanatory example of a two-dimensional discrete wavelet transform.

FIG. 4 is an explanatory example of mask information.

FIG. 5 is a peripheral area portion 32 which is a boundary with a designated area 30;
FIG.

FIG. 6 is an explanatory diagram of a designated area 30 and a peripheral area portion 32 after a subject has moved.

FIG. 7 is an explanatory example of the operation of the lower bit deletion device 18.

FIG. 8 is an example of a shift-up of a quantization index in a quantization device 20 (before a change).

FIG. 9 is an example (after change) of a quantization index shifted up in the quantization device 20;

FIG. 10 is an explanatory diagram of an operation of the entropy encoding device 22.

[Explanation of symbols]

 10: Image input device 12: Discrete wavelet transform device 14: Region designating device 16: Region surrounding calculation device 18: Lower bit removing device 20: Quantizer 22: Entropy coding device 24: Code output device 30: Designated region (ROI) (Area) 32: area peripheral area

Claims (8)

[Claims] 1. An area setting means for setting a predetermined area of an input image signal, and a boundary between the predetermined area and another area set by the area setting means is detected, and the area corresponding to the boundary is detected. A boundary portion detecting means for setting a boundary region; a compression encoding of an image signal of the predetermined region at a first compression ratio; and a video signal of the boundary region at a second compression ratio higher than the first compression ratio. An image compression encoding apparatus, comprising: encoding means for performing compression encoding and image encoding of the image signal in the other area at a third compression rate higher than the second compression rate. 2. An encoding device comprising: a sub-band encoding unit for performing sub-band encoding on the input image; and a lower bit value of output data in the boundary area among output data of the sub-band encoding unit. A lower bit reduction unit for reducing and outputting other data as it is, and quantizing output data of the lower bit reduction unit, and shifting up a quantization index in the boundary area among the quantization indexes obtained by the quantization. The image compression encoding apparatus according to claim 1, further comprising: a quantization unit; and an entropy encoding unit that entropy encodes an output of the quantization unit. 3. The area setting means sets a first area including the predetermined area and the boundary area, and the boundary area is inscribed in the first area set by the area setting means. The image compression encoding apparatus according to claim 1. 4. The image compression encoding apparatus according to claim 1, wherein the boundary area circumscribes the area set by the area setting means. 5. An area setting step of setting a predetermined area of an input image signal, and detecting a boundary between the predetermined area and another area set in the area setting step, and corresponding to the boundary. A boundary portion detecting step of setting a boundary region; and compressing and encoding the image signal of the predetermined region at a first compression ratio, and converting the image signal of the boundary region at a second compression ratio higher than the first compression ratio. An encoding step of compressing and encoding the image signal of the other area at a third compression ratio higher than the second compression ratio. 6. The sub-band encoding step for sub-band encoding the input image and the lower bit value of the output data in the boundary area among the output data of the sub-band encoding step. Reduce and
A lower bit reduction step of outputting other data as it is, and quantizing output data of the lower bit reduction step,
The method according to claim 5, further comprising: a quantization step of shifting up a quantization index in the boundary area among the quantization indexes obtained thereby; and an entropy encoding step of entropy encoding an output of the quantization means. Image compression encoding method as described in the above. 7. The area setting step sets a first area including the predetermined area and the boundary area, and the boundary area is inscribed in the first area set by the area setting means. An image compression encoding method according to claim 5. 8. The image compression encoding method according to claim 5, wherein the boundary area circumscribes the area set by the area setting means.