Digital image compression and transmission system visually weighted transform coefficients

• US 4,780,761 A
• Filed: 06/02/1987
• Issued: 10/25/1988
• Est. Priority Date: 06/02/1987
• Status: Expired due to Term

• Current Assignee: Eastman Kodak Company
• Original Assignee: Eastman Kodak Company
• Inventors: Daly, Scott J., Rabbani, Majid, Chen, Cheng-Tie
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US07/057,413
• Time in Patent Office: 511 Days
• Field of Search: 358/133, 358/138
• US Class Current: 375/240.2
• CPC Class Codes: H04N 19/60 using transform coding

Abstract

A system for compressing and transmitting a digital image signal over a limited bandwidth communication channel, transform codes the image values and quantizes the transform coefficients according to a two-dimensional model of the sensitivity of the human visual system. The model of the human visual system is characterized by being less sensitive to diagonally oriented spatial frequencies than to horizontally or vertically oriented spatial frequencies, thereby achieving increased compression of the image.

Claims

1. A transmitter for compressing and transmitting a digital image over a limited bandwidth communication channel comprising:

(a) means for performing a two-dimensional spatial frequency transformation on the digital image to produce transform coefficients; and

(b) means for quantizing the transform coefficients in accordance with a two-dimensional model of the sensitivity of the human visual system to spatial frequencies, said model of the human visual system including the feature that the human visual system is less sensitive to diagonally oriented spatial frequencies than to horizontally or vertically oriented spatial frequencies.

2. The transmitter claimed in claim 1, wherein said means for performing a spatial frequency transformation performs a discrete cosine transformation.

3. The transmitter claimed in claim 1, wherein said means for performing a spatial frequency transformation, performs said spatial frequency transformation on subportions of the digital image, and said model of the human visual system is further modified to have constant maximum response for spatial frequencies between a frequency where the response of the human visual system is a maximum and zero cycles per degree.

4. The transmitter claimed in claim 1, wherein said means for quantizing the transform coefficients comprises;

means for normalizing the transformed coefficients in accordance with said model of the human visual system; and

a uniform quantizer for quantizing the normalized coefficients.

5. The transmitter claimed in claim 4, wherein said means for normalizing the transform coefficients comprises;

a look-up table of normalizing coefficients and means for multiplying said transform coefficients by respective normalizing coefficients.

6. A receiver for receiving and decoding a coded, compressed digital image transmitted over a limited bandwidth communication channel, comprising:

(a) means for receiving an encoded digital image signal, the encoded digital image signal comprising code words representing transform coefficients of a spatial frequency transformation of a digital image, the transform coefficients being normalized in accordance with a two-dimensional model of the sensitivity of the human visual system, said model including the feature that the human visual system is less sensitive to diagonally oriented spatial frequencies than to horizontally or vertically oriented spatial frequencies; and

(b) means for decoding the encoded digital image signal, said decoding means including means for denormalizing the values of the transform coefficients, and means for inversely transforming the denormalized transform coefficients to recover the digital image signal.

7. A transceiver for compressing, transmitting and receiving a digital image over a limited bandwidth communication channel comprising:

(a) means for performing a two-dimensional spatial frequency transformation on a digital image to produce transform coefficients;

(b) means for normalizing said transform coefficients in accordance with a two-dimensional model of the sensitivity of the human visual system to spatial frequencies, said model of the human visual system including the features that the human visual system is less sensitive to diagonally oriented spatial frequencies than to horizontally or vertically oriented spatial frequencies;

(c) means for performing the inverse of said normalizing on said normalized transform coefficients to produce denormalized transform coefficients; and

(d) means for performing the inverse of said two-dimensional spatial frequency transformation on said denormalized transform coefficients to produce a digital image signal.

8. The transceiver claimed in claim 7, wherein said means for performing a spatial frequency transformation performs a discrete cosine transform.

9. The transceiver claimed in claim 8, wherein said means for quantizing the transform coefficients comprising:

means for normalizing the transformed coefficients in accordance with said model of the human visual system; and

a uniform quantizer for quantizing the normalized coefficients.

10. The transceiver claimed in claim 9, wherein said means for normalizing the transform coefficients comprises a look-up table of normalizing coefficients and means for multiplying said transform coefficients by respective normalizing coefficients.

11. The transceiver claimed in claim 10, wherein said means for performing a spatial transformation performs said spatial frequency transformation on subportions of the digital image, and said model of the human visual system is modified to have constant maximum response for spatial frequencies between a frequency where the response of the human visual system is maximum and zero cycles per degree.

12. A system for compressing and transmitting a digital image signal over a limited bandwidth communication channel and for receiving and displaying said digital image, said display means having a characteristic pixel spacing and viewing distance, comprising:

(a) transceiver means for transmitting and receiving said digital image, said transceiver means including,

(1) means for performing a two-dimensional spatial frequency transformation on a digital image to produce transform coefficients,

(2) means for normalizing said transform coefficients in accordance with a model of the human visual system,

(3) means for denormalizing the normalized transform coefficients to produce denormalized transform coefficients; and

(4) means for performing an inverse two-dimensional spatial frequency transformation on said denormalized transform coefficients to produce a digital image, said means for normalizing and means for denormalizing including a look-up table of normalization constants, and

(b) means for generating said look-up table of normalization constants including,

(1) means for determining the two-dimensional spatial frequency components of each of said transform coefficients,

(2) means for generating a two-dimensional model of the sensitivity of the human visual system to spatial frequencies, said model of the human visual system including the feature that the human visual system is less sensitive to diagonally oriented spatial frequencies than to horizontally or vertically oriented spatial frequencies, said model generating means including scaling means for receiving signal inputs representing said characteristic pixel spacing and viewing distance and scaling said model to said signal inputs, and

(3) means for identifying a visual sensitivity value with each transform coefficient to generate said normalization constants.

13. The system claimed in claim 12, wherein said means for performing a two-dimensional spatial frequency transformation performs a discrete cosine transformation.

14. The system claimed in claim 13, whereby said means for performing a two-dimensional spatial frequency transformation performs said spatial frequency transformation on subportions of the digital image and said means for generating the two-dimensional model of the sensitivity to the human visual system generates a model having constant maximum response for low frequencies and said means for receiving signal inputs also receives inputs representing the size of said subportion.