Image compression/decoding apparatus and method

• US 6,873,739 B2
• Filed: 05/14/2001
• Issued: 03/29/2005
• Est. Priority Date: 05/15/2000
• Status: Expired due to Fees

• Current Assignee: EMI (Entegris Incorporated)
• Original Assignee: EMI (Entegris Incorporated)
• Inventors: Lee, Seonyoung, Cho, Kyeongsoon, Hong, Seogil
• Primary Examiner(s): Do, Anh Hong
• Application Number:
  US09/855,864
• Publication Number:
  US 20020031275A1
• Time in Patent Office: 1,415 Days
• Field of Search: 382/240, 382/248, 382/233, 375/240.11
• US Class Current: 382/240
• CPC Class Codes: G06T 9/005 Statistical coding, e.g. Huffman, run length coding
• CPC Class Codes: G06T 9/007 Transform coding, e.g. discrete cosine transform

Abstract

An image compression apparatus is provided, which includes a wavelet transformer, a quantizer, a comparator, a subtracter, a low data compressor and a high data compressor. The comparator compares the first quantization coefficients of the first quantization data, to output the maximum quantization coefficient. The subtracter subtracts the minimum quantization coefficient from the maximum quantization coefficient, to output a subtraction constant. The low data compressor subtracts the subtraction constant from the first quantization coefficients, to output low compression data obtained by compressing the first quantization data. The high data compressor sequentially generates first compression data obtained by coding the number of zero quantization coefficients each having at least one successive zero value among the second quantization coefficients, and second compression data obtained by coding the second quantization coefficients other than the zero quantization coefficients according to a coding rule, to output high compression data.

Claims

1. An image compression apparatus comprising:

a wavelet transformer which receives digital image data, wavelet-transforms the digital image data in at least one step, and divides the digital image data into a first wavelet region having the lowest frequency and a second wavelet region having the other frequency, to thereby output image data of the first wavelet region and image data of the second wavelet region;

a quantizer which receives both the image data of the first and second wavelet regions and respectively quantizes the image data of the first and second wavelet regions to thereby output first quantization data formed of first quantization coefficients and second quantization data formed of second quantization coefficients, respectively;

a comparator which receives the first quantization data and compares the first quantization coefficients of the first quantization data, to thereby output the maximum quantization coefficient having the maximum quantization coefficient value and the minimum quantization coefficient having the minimum quantization coefficient value, among the first quantization coefficients;

a subtracter which receives the maximum quantization coefficient and the minimum quantization coefficient output from the comparator and subtracts the latter from the former, to thereby output a subtraction constant;

a low data compressor which receives the first quantization data and the subtraction constant and subtracts the subtraction constant from the first quantization coefficients of the first quantization data, to thereby output low compression data obtained by compressing the first quantization data; and

a high data compressor which sequentially receives the second quantization coefficients of the second quantization data and sequentially generates first compression data obtained by coding the number of zero quantization coefficients each having at least one successive zero value among the second quantization coefficients, and second compression data obtained by coding the second quantization coefficients other than the zero quantization coefficients according to a coding rule with reference to the coding rule where compression codes are established according to a respective quantization coefficient, to thereby output high compression data.

2. The image data compression apparatus of claim 1, wherein the first compression data is obtained by counting the number of zero quantization coefficients having at least one successive zero value among the second quantization coefficients and coding the successive zero quantization coefficients into the counted result.

4. The image data compression apparatus of claim 1, wherein the coding rule comprises coding data which is comprised of an upper bit group including a number of bits and a lower bit group including a number of bits, depending upon the second quantization coefficients,

wherein if an absolute value of the second quantization coefficient is one, first coding data of the upper bit group is 11 of a binary number,

wherein as the absolute value of the second quantization coefficient QUAN2 increases, zero of a binary number is inserted between 1 of a binary number and 1 of a binary number in the first coding data,

wherein the number into which zero of the binary number is inserted is the number of significant bits which are obtained by subtracting one of a decimal number from the absolute value of the second quantization coefficient and transforming the subtraction result into a binary number,

wherein if a second quantization coefficient is +1, the second coding data of the lower bit group is one of a binary number,

wherein if a second quantization coefficient is −

1, the third coding data of the lower bit group is zero of a binary number,

wherein if a quantization coefficient is a positive number and larger than +1, the coding data of the lower bit group contains bits obtained by transforming a decimal number obtained by {second quantization coefficient−

(2^{(number of significant bits−}

1)+1)}*2+1 into a binary number, and

wherein if a quantization coefficient is a negative number and smaller than −

1, the coding data of the lower bit group contains bits obtained by transforming a decimal number obtained by {absolute value of second quantization coefficient−

(2<sup>(number of significant bits−

1)</sup>+1)}*2 into a binary number.

3. The image data compression apparatus of claim 3, wherein the first compression data is inversely transformed from the uppermost bit of the counted binary number data to the lowermost bit thereof,

wherein the inversely transformed data is divided into two bits by two bits and then zero of a binary number is inserted every two bits,

wherein zero of a binary number is inserted in front of the uppermost bit of the inversely transformed data, and

wherein zero of a binary number is not inserted in the next of the lowermost bit of the inversely transformed data.

5. An image compression method comprising:

a wavelet transformation process which receives digital image data, wavelet-transforms the digital image data in at least one step, and divides the digital image data into a first wavelet region having the lowest frequency and a second wavelet region having the other frequency;

a quantization process which quantizes the image data of the first wavelet region and the image data of the second wavelet region to thereby generate first quantization data formed of first quantization coefficients and second quantization data formed of second quantization coefficients, respectively;

a quantization data judgement process which receives the first and second quantization data generated in the quantization process, and judges whether currently received quantization data is the first quantization data or the second quantization data;

a comparison process which compares the first quantization coefficients of the first quantization data, if it is judged that the currently received quantization data is the first quantization data in the quantization data judgement process, to thereby output the maximum quantization coefficient having the maximum quantization coefficient value and the minimum quantization coefficient having the minimum quantization coefficient value, among the first quantization coefficients;

a subtraction process which subtracts the minimum quantization coefficient from the maximum quantization coefficient, which are generated in the comparison process, to thereby output a subtraction constant;

a low data compression process which subtracts the subtraction constant from the first quantization coefficients of the first quantization data, to thereby output low compression data obtained by compressing the first quantization data; and

a high data compression process which codes the number of zero quantization coefficients each having at least one successive zero value among the sequentially input second quantization coefficients, if the received quantization data is the second quantization data in the quantization data judgment process, and codes the second quantization coefficients other than the zero quantization coefficients according to a coding rule with reference to the coding rule where compression codes are established according to a respective quantization coefficient, to thereby output high compression data.

6. The image data compression method of claim 5, further comprising a compression data selection process which receives a region select signal for distinguishing the first wavelet region from the second wavelet region, the low compression data and the high compression data and selects one of the low compression data and the high compression data according to the region select signal.

7. The image data compression method of claim 5, wherein the high data compression process comprises:

a zero coefficient judgment step of judging whether sequentially input second quantization coefficients are zero quantization coefficients each having a zero value;

a counting step of counting the number of the zero quantization coefficients if the second quantization coefficients are zero quantization coefficients in the zero coefficient judgment step;

a first compression step of coding the counted result of the counting step into data of a binary number to then generate first compression data, and reset the counted number of the counting step; and

a second compression step of coding the second quantization coefficients according to a coding rule with reference to the coding rule where compression codes are established according to a respective quantization coefficient if the second quantization coefficients are not the zero quantization coefficients in the zero coefficient judgment step.

8. The image data compression method of claim 5, wherein coding of the zero quantization coefficients in the high data compression process is performed by coding zero quantization coefficients into zero of a binary number.

9. The image data compression method of claim 7 wherein the binary number data counted in the counting process in the case of the first compression data of the first compression process is inversely transformed from the uppermost bit to the lowermost bit,

wherein the inversely transformed data is divided into two bits by two bits, and then zero of a binary number is inserted every two bits,

wherein zero of a binary number is inserted in front of the uppermost bit of the inversely transformed data, and

wherein zero of a binary number is not inserted in the next of the lowermost bit of the inversely transformed data.

10. The image data compression method of claim 5, wherein the coding rule comprises coding data which is comprised of an upper bit group including a number of bits and a lower bit group including a number of bits, depending upon the second quantization coefficients,

wherein if an absolute value of the second quantization coefficient is one, first coding data of the upper bit group is 11 of a binary number,

wherein as the absolute value of the second quantization coefficient QUAN2 increases, zero of a binary number is inserted between 1 of a binary number and 1 of a binary number in the first coding data,

wherein the number into which zero of the binary number is inserted is the number of significant bits which are obtained by subtracting one of a decimal number from the absolute value of the second quantization coefficient and transforming the subtraction result into a binary number,

wherein if a second quantization coefficient is +1, the second coding data of the lower bit group is one of a binary number,

wherein if a second quantization coefficient is −

1, the third coding data of the lower bit group is zero of a binary number,

wherein if a quantization coefficient is a positive number and larger than +1, the coding data of the lower bit group contains bits obtained by transforming a decimal number obtained by {second quantization coefficient−

(2<sup>(number of significant bits−

1)</sup>+1)}*2+1 into a binary number, and

wherein if a quantization coefficient is a negative number and smaller than −

1, the coding data of the lower bit group contains bits obtained by transforming a decimal number obtained by {absolute value of second quantization coefficient−

(2^{(number of significant bits−}

1)+1)}*2 into a binary number.

11. An image data compression method comprising:

a wavelet transformation process of receiving digital image data, and wavelet-transforming the digital image data in at least one step;

a quantization process of quantizing the wavelet transformed image data to thereby generate quantization data;

a zero coefficient judgement process of judging whether the quantization coefficients of the sequentially input quantization data are zeros;

a counting process of counting the number of the zero quantization coefficients if the quantization coefficients are zeros in the zero coefficient judgment process;

a first compression process of coding the counted result of the counting step into data of a binary number to then generate first compression data, and reset the counted number of the counting process; and

a second compression process of coding the quantization coefficients according to a coding rule with reference to the coding rule where compression codes are established according to a respective quantization coefficient if the quantization coefficients are not zero in the zero coefficient judgment process.

12. The image data compression method of claim 11, wherein the binary number data counted in the counting process in the case of the first compression data of the first compression process is inversely transformed from the uppermost bit to the lowermost bit,

wherein the counted number of the counting process is divided into two bits by two bits,

wherein zero of a binary number is inserted every two bits, and

wherein zero of a binary number is not inserted in the next of the lowermost two bits of the counted number.

13. The image data compression method of claim 11, wherein the coding rule comprises coding data which is comprised of an upper bit group including a number of bits and a lower bit group including a number of bits, depending upon the second quantization coefficients,

wherein if an absolute value of the second quantization coefficient is one, first coding data of the upper bit group is 11 of a binary number,

wherein as the absolute value of the second quantization coefficient QUAN2 increases, zero of a binary number is inserted between 1 of a binary number and 1 of a binary number in the first coding data,

wherein the number into which zero of the binary number is inserted is the number of significant bits which are obtained by subtracting one of a decimal number from the absolute value of the second quantization coefficient and transforming the subtraction result into a binary number,

wherein if a second quantization coefficient is +1, the second coding data of the lower bit group is one of a binary number,

wherein if a second quantization coefficient is −

1, the third coding data of the lower bit group is zero of a binary number,

wherein if a quantization coefficient is a positive number and larger than +1, the coding data of the lower bit group contains bits obtained by transforming a decimal number obtained by {second quantization coefficient−

(2<sup>(number of significant bits−

1)</sup>+1)}*2+1 into a binary number, and

wherein if a quantization coefficient is a negative number and smaller than −

1, the coding data of the lower bit group contains bits obtained by transforming a decimal number obtained by {absolute value of second quantization coefficient−

(2<sup>(number of significant bits−

1)</sup>+1)}*2 into a binary number.

14. An image data compression method comprising:

a wavelet transformation process of receiving digital image data, and wavelet-transforming the digital image data in at least one step;

a quantization process of quantizing the wavelet transformed image data to thereby generate quantization data;

a comparison process of generating the maximum quantization coefficient having the largest quantization coefficient value and the minimum quantization coefficient having the smallest quantization coefficient value, among the quantization data;

a subtraction process of subtracting the minimum quantization coefficient from the maximum quantization coefficient generated in the comparison process, to thereby generate a subtraction constant; and

a compression process of subtracting the subtraction constant generated in the subtraction process from each of the quantization coefficients in the quantization data, to thereby compress the quantization data.

15. An image decoding apparatus comprising:

a first judgment unit which receives image compression data and judges whether the received image compression data is low compression data or high compression data, to thereby output the low compression data and the high compression data;

an adder which receives the low compression data of the first judgment unit and adds a subtraction constant to the received low compression data and transforms the received low compression data into first quantization data;

a second judgment unit which receives the high compression data of the first judgment unit and judges whether the received high compression data is first compression data or second compression data, to thereby output the first and second compression data;

a first decoder which receives the first compression data of the second judgment unit and outputs zero quantization coefficients each having a zero value;

a second decoder which receives the second compression data of the second judgment unit and outputs second quantization data according to a coding rule with reference to the coding rule where quantization coefficients are established depending upon compression codes;

an inverse quantizer which receives the first quantization data, the zero quantization data and the second quantization data, inversely quantizes the first quantization data to then output first wavelet region image data having the lowest frequency, inversely quantizes the zero quantization data and the second quantization data to then second wavelet region image data having frequencies other than the lowest frequency; and

an inverse wavelet transformer which receives the first wavelet region image data and the second wavelet region image data and inversely wavelet-transforms the received first and second wavelet region image data, to thereby output decoded digital image data.

16. The image decoding apparatus of claim 15, wherein the zero quantization data is formed of zero quantization coefficients each having a zero value as many as a corresponding value depending upon the value of the first compression data.

17. The image decoding apparatus of claim 15, wherein the zero quantization data is formed of zero quantization coefficients each having a zero value as many as a corresponding value depending upon a value corresponding to an inversely transformed data value obtained by dividing the first compression data from the uppermost bit three bits by three bits, and inversely transforming data composed of remaining bits excluding the first bit among three bits from the uppermost bit to the lowermost bit.

18. An image decoding method comprising:

a first judgment process of receiving image compression data and judges whether the received image compression data is low compression data or high compression data;

an adding process of adding a subtraction constant to the low compression data of the first judgment process and transforming the low compression data into first quantization data;

a second judgment process of judging whether the high compression data of the first judgment process is first compression data or second compression data;

a first decoding process of receiving the first compression data of the second judgment process and outputting zero quantization data each having a zero value;

a second decoding process of receiving the second compression data of the second judgment process and outputting second quantization data according to a coding rule with reference to the coding rule where quantization coefficients are established depending upon compression codes;

an inverse quantization process of receiving the first quantization data, the zero quantization data and the second quantization data, inversely quantizing the first quantization data to then output first wavelet region image data having the lowest frequency, inversely quantizing the zero quantization data and the second quantization data to then second wavelet region image data having frequencies other than the lowest frequency; and

an inverse wavelet transformation process of receiving the first wavelet region image data and the second wavelet region image data and inversely wavelet-transforming the received first and second wavelet region image data, to thereby output decoded digital image data.