JP2795171B2 - Image compression method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to compression of image data, and more particularly, to discrete cosine transform (DCT).
The present invention relates to a method and an apparatus for performing efficient image compression using orthogonal transform such as ransform.

[0002]

2. Description of the Related Art In general, when a video signal is converted from analog to digital, the amount of information is greatly increased. Therefore, television broadcasting, image communication, image recording to a storage medium, and the like are performed at a practical level as it is. Can not do. Therefore, several international standard methods for compressing image data have been proposed, and various image compressing apparatuses have been proposed.

FIG. 3 is a block diagram showing a basic configuration of a conventional image compression apparatus. The image data is stored in the memory 101, but is rearranged by the memory control unit 102 in macroblock units. The macro block is made up of a total of 6 blocks of 4 blocks of a luminance signal and 2 blocks of a color difference signal, and each block is composed of 8 × 8 pixels.

[0004] Data read out from the memory 101 in block units is orthogonally transformed into a frequency series by the DCT converter 103, thereby increasing the low-frequency component.
Quantization of the CT output coefficients by the quantizer 104 reduces spatial redundancy of image data. The data is subjected to entropy encoding by the variable-length encoder 105, whereby compressed encoded data can be obtained.

[0005] In addition to such a basic structure, an image coding apparatus has been proposed in which a low-pass filter is provided in the preceding stage of the DCT converter, and the pass band is changed according to the transmission rate or the like to improve the image compression efficiency. (Japanese Patent Laid-Open No. 4-248)
787). This image encoding device utilizes a characteristic that the DCT transform output of a pixel block is concentrated in a low-frequency portion of a spatial frequency, and applies a low-pass filter to a video signal before DCT transform to reduce high-frequency components in advance. Thus, the compression efficiency is improved.

In another conventional example, there is disclosed an encoding apparatus which divides an input image signal into four sub-bands and attenuates a high-frequency component of each band signal to adjust the amount of encoded data. (JP-A-3-209988).

[0007]

However, in the above-mentioned conventional image coding apparatus, the quantization step size and low-pass size are determined according to only the conditions on the coded data output side such as the transmission rate, the frame rate, and the occupation state of the output buffer memory. This is to change the pass band of the filter. Similarly, the encoding apparatus of the subband division system switches the filter according to the data amount of the buffer memory, and adjusts the encoded data amount so as to match the transmission rate.

Therefore, in the above-mentioned conventional technique, the input image is only band-limited uniformly according to the conditions on the output side, and efficient compression suitable for the input image is not performed.
For this reason, it has not been possible to achieve high image quality while efficiently suppressing the amount of generated code data.

An object of the present invention is to provide an image compression method and apparatus capable of obtaining high image quality by efficient image compression.

[0010]

An image compression method according to the present invention utilizes an orthogonal transform for converting a time-series signal into a frequency-sequence coefficient. The frequency-sequence coefficient obtained by orthogonally transforming input image data is used. And a filter band control step of controlling the pass band of the filter unit in accordance with the upper limit value, and using orthogonal transform on the image data that has passed through the filter unit whose filter band has been controlled. And an image compression encoding step for performing image compression.

[0011] Preferably, the filter band control step includes a step of orthogonally transforming the input image data to generate a frequency sequence coefficient, and detecting upper limit values of horizontal, vertical and oblique frequency components using the frequency sequence coefficient. And controlling each of the horizontal, vertical, and oblique pass bands of the filter unit according to each upper limit value.

The orthogonal transform used in the present invention is preferably a discrete cosine transform (DCT). In this case, the filter band control step performs DCT conversion of the input image
Generating a T output coefficient; detecting an upper limit value of a frequency component based on the DCT output coefficient; and controlling a pass band of the input image data according to the upper limit value of the frequency component.

The control of the filter passband is preferably
This is performed in such a manner that the cutoff frequency of the filter unit matches the detected upper limit value.

An image compression apparatus according to the present invention has DCT transform means for generating DCT output coefficients by subjecting input image data to DCT in units of macroblocks, and filter means for changing a pass band of the input image data. When,
Filter control means for detecting the upper limit value of the frequency component in each macro block using the DCT output coefficient and controlling the pass band of the filter means in accordance with the upper limit value, and a macro block passing through the filter means controlled by the filter control means And image compression encoding means for executing image compression using DCT transformation on the image data.

[0015]

First, image data is transformed by orthogonal transformation (or DCT).
Conversion), the frequency distribution of the image data is analyzed using the frequency sequence coefficients obtained thereby, and the pass band of the filter unit is controlled according to the upper limit value of the frequency component. The control of the pass band is performed in such a manner that the pass band of the filter is limited to the low band when the high frequency component is small, and the pass band of the filter is extended to the high band when the high frequency component is large. In this way, the band of the image data is limited through the filter unit whose pass band is controlled, and compression-encoding processing including orthogonal transform is performed on the band-limited image data.

[0016]

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

FIG. 1 is a block diagram showing an embodiment of an image compression apparatus according to the present invention. In FIG.
The input and output are controlled by the memory controller 2, and the input image data is rearranged in units of macroblocks under the control of the memory controller 2 and stored in the memory 1.

The filter section includes a horizontal low-pass filter 3 capable of changing the horizontal band within the macroblock, a vertical low-pass filter 4 capable of similarly changing the vertical band, and a diagonal band. It has a configuration in which oblique low-pass filters 5 that can be changed are connected in cascade.

The image data that has passed through the filter unit is DC
DCT conversion is performed by the T converter 6 in block units, and D
The CT output coefficient is output to the filter control unit 7 and the quantizer 8. Low-pass filters 3, 4, and 5 of filter section
Are controlled by the filter control unit 7 as described later, and their cutoff frequencies are set. The image data that has passed through the filter unit whose cutoff frequency is controlled is subjected to DCT transformation again, quantized by the quantizer 8, entropy-encoded by the variable-length encoder 9, and output as encoded data.

The DCT converter 6 outputs an 8 × 8 DCT output coefficient block for the 8 × 8 pixel block, and the DCT output coefficient represents the distribution of frequency components in the pixel block. Upon input of the DCT output coefficient block, the filter control unit 7 detects the upper limit value of the frequency component in each of the horizontal, vertical, and diagonal directions. That is, a threshold determined for each frequency in each direction is used, and the maximum frequency component having an amplitude equal to or greater than the threshold is detected as the upper limit. The filter control unit 7 controls the horizontal, vertical, and oblique low-pass filters 3, 4, and 5.
, Respectively, to make their cutoff frequencies coincide with the respective detected upper limit values. In other words, control is performed so that the band of the filter is widened for data with a wide band, and the band of the filter is narrowed for data with a narrow band.

FIG. 2 is a schematic flowchart for explaining one embodiment of the image compression method according to the present invention, and explains the operation of the apparatus shown in FIG. As shown in FIG. 2, the image compression method according to the present invention includes a band control routine of the filter unit and an actual image compression routine.

Filter Band Control First, image data is input, rearranged in macroblock units, and stored in the memory 1 (S10). continue,
A macroblock MBi (i is an integer equal to or larger than 1 and equal to or smaller than the number of macroblocks of one screen) is read from the memory 1 through the filter unit (S11), and the DCT converter 6
, And is output to the filter control unit 7 as a DCT output coefficient block. However, at this time, the low-pass filters 3, 4, and 5 of the filter unit do not perform band limitation, and all bands of the image data pass through the DCT.
Input to the converter 6.

The filter control unit 7 receives the DCT output coefficient block, detects the upper limit values of the frequency components in the horizontal, vertical and oblique directions as described above (S13), and these upper limit values become the cutoff frequencies. The low-pass filters 3, 4 and 5 are controlled as described above (S14). This makes it possible to control the horizontal, vertical, and oblique bands of the image in accordance with the frequency distribution of the read macro blocks MBi. That is, when the high-frequency component is small, the cutoff frequency of the filter is lowered to narrow the band, and when the high-frequency component is large, the cutoff frequency is increased to widen the band, thereby efficiently generating the encoded data. It is possible to suppress the image quality and to perform compression encoding without deteriorating the image quality.

The macroblock MB is determined by the image compression / encoding filter band control routine.
When the cutoff frequency for i is set, the macro block MBi is read out again from the memory 1 (S15) and sent to the quantizer 8 and the variable length encoder 9 (S16).
Output as encoded data.

When the filter band control for the macroblock MBi is completed, i is incremented (S17), and the next macroblock MBi + 1 is read from the memory 1 (S1).
1). Then, filter band control is performed on the macro block MBi + 1 (S12 to S14), and similarly,
Filter band control and compression coding are repeated for all macroblocks.

[0026]

As described above in detail, the image compression method and apparatus according to the present invention perform orthogonal transformation (or DCT transformation) on image data prior to actual compression encoding.
The pass band of the filter unit is limited based on the frequency sequence coefficient obtained thereby. After that, the image data passes through the filter section whose pass band is limited, and compression encoding processing including orthogonal transform is performed. For this reason, the band limitation most suitable for the image data to be compressed is performed, the amount of data to be encoded can be optimized, and high-quality compression encoding can be achieved.

[Brief description of the drawings]

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

FIG. 2 is a schematic flowchart illustrating an embodiment of an image compression method according to the present invention.

FIG. 3 is a block diagram illustrating a basic configuration of a conventional image compression device.

[Explanation of symbols]

 Reference Signs List 1 memory 2 memory control unit 3 horizontal low-pass filter 4 vertical low-pass filter 5 diagonal low-pass filter 6 discrete cosine transform (DCT) unit 7 filter control unit 8 quantizer 9 variable-length encoder 10

Claims (12)

(57) [Claims] 1. An image compression method using an orthogonal transform for converting a time-series signal into a frequency-sequence coefficient, wherein an upper limit value of a frequency component is detected using a frequency-sequence coefficient obtained by performing the orthogonal transform on the input image data. A filter band control step of changing the pass band of the filter unit so as to increase the pass band of the filter unit as the upper limit increases and to reduce the pass band of the filter unit as the upper limit decreases. An image compression encoding step of performing image compression using the orthogonal transform on the input image data that has passed through the filter unit that has been subjected to the filter band control. 2. The image compression method according to claim 1, wherein in the filter band control step, a cutoff frequency of the filter unit is made to match the upper limit. 3. An image compression method using an orthogonal transform for converting a time-series signal into a frequency sequence coefficient, comprising: a first step of orthogonally transforming input image data to generate the frequency sequence coefficient; A second step of detecting upper limit values of frequency components in the horizontal, vertical and diagonal directions, respectively, and as the pass band in the horizontal, vertical and diagonal directions of the filter unit increases as the upper limit value increases and decreases as the upper limit value decreases. A third step of controlling the horizontal, vertical and oblique passbands of the filter unit so as to reduce the size, and using the orthogonal transformation on the input image data that has passed through the passband-limited filter unit. A fourth step of performing image compression; and an image compression method. 4. The image compression method according to claim 3, wherein in the third step, the cutoff frequencies in the horizontal, vertical, and oblique directions of the filter unit are made to match the respective upper limits. 5. An image compression method using a discrete cosine transform (DCT), comprising: a step of storing input image data in macroblock units; a step of performing DCT conversion on the stored image data; Generating an upper limit value of the frequency component of the macroblock using the DCT output coefficient corresponding to an arbitrary macroblock, and expanding the pass band of the filter unit as the upper limit value increases. A filter band control step of controlling the pass band of the filter unit so that the pass band of the filter unit is reduced as the value becomes lower. The DCT transform is performed on the macroblock that has passed through the filter unit whose filter band is controlled. And an image compression encoding step for performing the image compression used. Image compression method to be. 6. The filter band limiting step includes: DCT transforming the macroblock to generate a DCT output coefficient; and determining horizontal, vertical, and diagonal frequency components of the macroblock based on the DCT output coefficient. Detecting the upper limit value, and the horizontal, vertical, and diagonal directions of the filter unit, so that the horizontal, vertical, and diagonal passbands of the filter unit expand and decrease as the respective upper limit values increase and decrease. 6. The image compression method according to claim 5, further comprising: controlling a pass band. 7. The image compression method according to claim 5, wherein in the filter band control step, a cutoff frequency of the filter unit is made to match the upper limit. 8. An image compression method using discrete cosine transform (DCT), a first step of storing input image data in macroblock units, and a second step of reading out the stored image data in macroblock units A third step of performing a DCT transform on the read image data of the macroblock to generate a DCT output coefficient, and a fourth step of detecting an upper limit value of a frequency component in the macroblock using the DCT output coefficient A fifth step of limiting the passband of the image data of the macroblock so that the passband of the image data of the macroblock is increased as the upper limit is increased and is reduced as the upper limit is decreased. Image compression using DCT transform for band-limited macroblock image data 6 and step consists, image compression method characterized by repeated up to the sixth step from the second step in macroblock units to run. 9. The method according to claim 9, wherein the fourth step includes detecting upper limit values of horizontal, vertical, and diagonal frequency components of the macroblock using the DCT output coefficients. Limiting the horizontal, vertical and diagonal passbands of the macroblock so that the horizontal, vertical and diagonal passbands of the block are enlarged as the respective upper limit values are higher and are reduced as the upper limit values are lower. The image compression method according to claim 8, wherein: 10. An image compression apparatus having DCT transform means for performing discrete cosine transform (DCT) on input image data in macroblock units to generate DCT output coefficients, wherein a pass band of the input image data can be changed. Filter means, using the DCT output coefficient to detect an upper limit value of a frequency component in each macroblock, so that the higher the upper limit value is, the larger the pass band of the filter means is, and the lower the upper limit value is, the smaller the lower limit value is, Filter control means for controlling a pass band of the filter means, and an image compression code for performing image compression using the DCT transform on image data of a macroblock passing through the filter means controlled by the filter control means An image compression device comprising: an image compression unit. 11. An apparatus for compressing and encoding input image data by discrete cosine transform (DCT), quantization, and entropy encoding, wherein: memory means for rearranging and storing the input image data in macroblock units; Reading means for reading the stored image data from the memory means in macroblock units; filter means for changing a pass band of the read image data; and reading from the memory means in macroblock units. DCT transformation means for transforming the image data into DCT to generate a DCT output coefficient, detecting an upper limit value of a frequency component in each macroblock using the DCT output coefficient, and passing the filter means as the upper limit value increases. The filter is expanded so that the band is expanded and the lower the upper limit becomes, the smaller the band becomes. Filter control means for controlling a pass band of the means, and image compression coding means for performing compression coding on image data of a macroblock passing through the filter means controlled by the filter control means. An image compression device characterized by the above-mentioned. 12. The filter means comprises: a horizontal low-pass filter capable of changing a horizontal pass band in the macro block; and a vertical low-pass filter capable of changing a vertical pass band in the macro block. And a diagonal low-pass filter capable of changing a passband in a diagonal direction within the macroblock. The filter control means uses the DCT output coefficient to calculate the horizontal, vertical and diagonal directions of the macroblock. 12. The image according to claim 10, wherein an upper limit value of each frequency component is detected, and the cutoff frequencies of the horizontal, vertical, and diagonal low-pass filters are controlled to match respective upper limit values in a corresponding direction. Compression device.