Method and apparatus for compressing and expanding image data

• US 6,661,924 B1
• Filed: 09/07/2000
• Issued: 12/09/2003
• Est. Priority Date: 09/10/1999
• Status: Expired due to Fees

• Current Assignee: PENTAX RICOH IMAGING COMPANY, LTD.
• Original Assignee: Pentax Corporation (Hoya Corporation)
• Inventors: Abe, Nobuaki
• Primary Examiner(s): Johns, Andrew W.
• Assistant Examiner(s): Alavi, Amir
• Application Number:
  US09/657,103
• Time in Patent Office: 1,188 Days
• Field of Search: 382/184, 382/251, 382/233, 382/232, 382/287, 382/250, 382/295, 382/248, 382/296, 382/239, 382/297, 382/245, 382/299, 382/246, 382/298, 382/300, 345/648, 345/659, 345/660, 345/671, 345/672, 345/677, 348/449, 348/561, 348/581, 348/583, 348/704, 358/428, 358/451, 358/488, 358/525, 708/203, 708/208, 708/290, 250/557, 375/240.01, 375/240.02, 375/240.03, 375/240.06, 375/240.2, 375/240.23, 375/240.24, 375/240.25
• US Class Current: 382/232
• CPC Class Codes: H04N 19/33 in the spatial domain
• CPC Class Codes: H04N 19/59 involving spatial sub-sampling or interpolation, e.g. alteration of picture size or resolution
• CPC Class Codes: H04N 19/91 Entropy coding, e.g. variable length coding [VLC] or arithmetic coding

Abstract

An image compression apparatus transforms original image data partitioned into first blocks, each of which is composed of a plurality of pixels, to reduced-image data composed of a smaller number of pixels than that of the original image data. Further, the apparatus generates expanded-image data from the reduced-image data by fluency transform. Based on the expanded-image data and the original image data, differential value data is obtained, and the differential value is transformed to differential DCT coefficient data by DCT processing. Based on the differential DCT coefficient data, a code-length corresponding to a bit-length necessary for a Huffman coding is calculated. The fluency transform has a plurality of modes, each mode being selected in order, the code-length being calculated for all of the modes. Then, a mode which makes the code-length minimum is determined as the optimum mode. The differential DCT coefficient data is Huffman encoded, so that Huffman-encoded bit data is generated. Then, in an image expansion apparatus, the original image data is restored on using the optimum mode, reduced-image data and the Huffman-encoded bit data.

Claims

1. A compression apparatus for compressing image data comprising:

a reduced-image generating processor that transforms original image data partitioned into first blocks, each of which is composed of a plurality of pixels, to reduced-image data composed of a smaller number of pixels than that of said original image data;

a fluency transform processor that applies a fluency transform to said reduced-image data so as to generate expanded-image data partitioned into second blocks corresponding to said first blocks, said fluency transform having a plurality of modes;

a differential value calculating processor that obtains differential value data indicating a difference between said original image data and said expanded-image data;

an orthogonal transform processor that obtains orthogonal transform coefficient data by applying an orthogonal transform to said differential value data;

a mode setting processor that selects one mode from said plurality of modes, said orthogonal transform coefficient data being generated in accordance with said selected mode;

a code-length calculating processor that calculates a code-length corresponding to a bit length of Entropy-encoded bit data obtained by an Entropy coding to said orthogonal transform coefficient data, said code-length calculating processor calculating said code-length in each of said plurality of modes;

an optimum mode determining processor that determines an optimum mode, by which said code-length becomes minimum, from said plurality of modes; and

an optimum Entropy coding processor that obtains said Entropy-encoded bit data by applying said Entropy coding to said orthogonal transform coefficient data in accordance with said optimum mode.

2. The compression apparatus according to claim 1, further comprising a recording medium, said reduced-image data, said Entropy-encoded bit data and said optimum mode being recorded in said recording medium.

3. The compression apparatus according to claim 1, wherein each of said first blocks and each of said second blocks are composed of 8×

8 (=64) pixels respectively.

4. The compression apparatus according to claim 1, wherein said reduced-image generating processor generates said reduced-image data composed of average pixels, obtained by finding an average value of said plurality of pixels in each of said first blocks.

5. The compression apparatus according to claim 1, wherein said fluency transform processor generates said expanded-image data by applying the fluency transform, which is expressed by the following formula, to each pixel of said reduced-image data $f$

(t)=(1/2

π

)

∑

-π

π



 



F

(u)

 

[ 

m

o]

φ



(t,u)



u

note that, ${\hspace{0.17em}<br><br>}_{\left[{\hspace{0.17em}<br><br>}^m<br><br>o\right]}<br><br>\mathrm{\phi <br><br>}<br><br>\left(t,u\right)=\left(1/2<br><br>\mathrm{\pi <br><br>}\right)<br><br>{\int <br><br>}_{-\mathrm{\infty <br><br>}}^{\mathrm{\infty <br><br>}}<br><br>\left\{\underset{p=-\mathrm{\infty <br><br>}}{\overset{\mathrm{\infty <br><br>}}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{\delta <br><br>}<br><br>\left(\mathrm{\omega <br><br>}-u+2<br><br>\mathrm{\pi <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>p\right)\right\}\times <br><br>{\left\{\underset{q=-\mathrm{\infty <br><br>}}{\overset{\mathrm{\infty <br><br>}}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\left(1-q<br><br>\left(2<br><br>\mathrm{\pi <br><br>}/\mathrm{\omega <br><br>}\right)\right)}^{-2<br><br>m}\right\}}^{-1/2}<br><br>\text{\hspace{1em}<br><br>}\times <br><br>\exp <br><br>\text{\hspace{1em}<br><br>}<br><br>\left(\mathrm{<br><br>\omega <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>t\right)<br><br><br><br>\mathrm{\omega <br><br>}$

where F(u) corresponds to each pixel value of said reduced-image data, f(t) is output values of the fluency transform, φ

(t,u) is the fluency function defined by a fluency function space ^mS, and m (=1,2,3, . . . ) is a parameter indicating a differentiability, said parameter m corresponding to said plurality of modes.

6. The compression apparatus according to claim 5, wherein said fluency transform processor applies the fluency transform in a state such that each pixel of said reduced-image data are arranged at a center position of each of said second blocks, and generates said expanded-image data composed of said plurality of pixels by adding said output values f(t) corresponding to pixel positions in said second blocks.

7. The compression apparatus according to claim 6, wherein said plurality of pixels in said original image data is arranged in a matrix, said fluency transform processor applies the fluency transform along a width direction to each pixel of said reduced-image data so that pixels aligned along the width direction are generated, and then applies the fluency transform along a length direction to said pixels aligned along said width direction such that said plurality of pixels in each of said second blocks is generated.

8. The compression apparatus according to claim 1, wherein said optimum mode is determined in each of said second blocks.

9. The compression apparatus according to claim 1, wherein said orthogonal transform processor applies a DCT (Discrete Cosine Transform) processing to said differential value data such that a differential DCT coefficient data is obtained.

10. The compression apparatus according to claim 1, wherein said Entropy coding processor applies a Huffman coding to said orthogonal transform coefficient data so that Huffman-encoded bit data is obtained, said code-length calculating processor calculating said code-length corresponding to said Huffman-encoded bit data, said optimum mode determining processor determining a mode, which makes said code-length minimum, among said plurality of modes as said optimum mode.

11. An expansion apparatus for expanding said reduced-image data recorded in said recording medium by said compression apparatus in claim 2, said expansion apparatus comprising:

a data reading processor that reads said reduced-image data, said Entropy-encoded bit data and said optimum mode recorded in said recording medium;

an optimum mode setting processor that sets said optimum mode among said plurality of modes;

an expanded-image generating processor that applies said fluency transform according to said optimum mode to said reduced-image data such that said expanded-image data is obtained;

an Entropy decoding processor that restores said orthogonal transform coefficient data by applying an Entropy-decoding to said Entropy-encoded bit data;

an inverse orthogonal transform processor that restores said differential value data by applying an inverse orthogonal transform to said orthogonal transform coefficient data;

an original image data restoring processor that restores said original image data on the basis of said expanded-image data and said differential value data.

12. The expansion apparatus according to claim 11, wherein said Entropy-encoded bit data is Huffman-encoded bit data, said Entropy decoding processor applying a Huffman decoding to said Huffman-encoded bit data.

13. The expansion apparatus according to claim 11, wherein said orthogonal transform coefficient data is a differential DCT coefficient data, said inverse orthogonal transform processor applying an inverse DCT processing to said differential DCT coefficient data.

14. A compression method for compressing image data comprising:

transforming original image data partitioned into first blocks, each of which is composed of a plurality of pixels, to reduced-image data composed of a smaller number of pixels than that of said original image data;

applying a fluency transform to said reduced-image data so as to generate expanded-image data partitioned into second blocks corresponding to said first blocks, said fluency transform having a plurality of modes;

obtaining differential value data indicating a difference between said original image data and said expanded-image data;

obtaining orthogonal transform coefficient data by applying an orthogonal transform to said differential value data;

selecting one mode from said plurality of modes, said orthogonal transform coefficient data being generated in accordance with said selected mode;

calculating a code-length corresponding to a bit length of Entropy-encoded bit data obtained by an Entropy coding to said orthogonal transform coefficient data, said code-length being calculated in each of said plurality of modes;

determining an optimum mode, by which said code-length becomes minimum, among said plurality of modes; and

obtaining said Entropy-encoded bit data by applying said Entropy coding based on said optimum mode to said orthogonal transform coefficient data.

15. An expansion method for expanding said reduced-image data obtained by said compression method in claim 14, said expansion method comprising:

reading said reduced-image data, said Entropy-encoded bit data and said optimum mode recorded in a recording medium;

setting said optimum mode among said plurality of modes;

applying said fluency transform according to said optimum mode to said reduced-image data such that said expanded-image data is obtained;

restoring said orthogonal transform coefficient data by applying an Entropy-decoding to said Entropy-encoded bit data;

restoring said differential value data by applying an inverse orthogonal transform to said orthogonal transform coefficient data;

restoring said original image data on the basis of said expanded-image data and said differential value data.

16. A memory medium that stores a program for compressing image data, said program comprising:

transforming original image data partitioned into first blocks, each of which is composed of a plurality of pixels, to reduced-image data composed of a smaller number of pixels than that of said original image data;

applying a fluency transform to said reduced-image data so as to generate expanded-image data partitioned into second blocks corresponding to said first blocks, said fluency transform having a plurality of modes;

obtaining differential value data indicating a difference between said original image data and said expanded-image data;

obtaining orthogonal transform coefficient data by applying an orthogonal transform to said differential value data;

selecting one mode from said plurality of modes, said orthogonal transform coefficient data being generated in accordance with said selected mode;

calculating a code-length corresponding to a bit length of an Entropy-encoded bit data obtained by an Entropy coding to said orthogonal transform coefficient data, said code-length being calculated in each of said plurality of modes;

determining an optimum mode, by which said code-length becomes minimum, among said plurality of modes; and

obtaining said Entropy-encoded bit data by applying said Entropy coding, based on said optimum mode, to said orthogonal transform coefficient data.

17. A memory medium that stores a program for expanding said reduced-image data obtained by said program for compressing image data in claim 16, said program for expanding comprising:

reading said reduced-image data, said Entropy-encoded bit data and said optimum mode recorded in a recording medium;

setting said optimum mode from said plurality of modes;

applying said fluency transform according to said optimum mode to said reduced-image data such that said expanded-image data is obtained;

restoring said orthogonal transform coefficient data by applying an Entropy-decoding to said Entropy-encoded bit data;

restoring said differential value data by applying an inverse orthogonal transform to said orthogonal transform coefficient data;

restoring said original image data on the basis of said expanded-image data and said differential value data.