Method and system for low-subband content discrimination

1. A method of discriminating noise from content when compressing digital video data, comprising:

computing a motion compensated difference frame from current image data and a reference frame;

applying an orthogonal transform to the motion compensated difference frame in a pixel domain to obtain a transformed motion compensated difference frame having transform coefficients in a frequency domain, the transform coefficients including a DC component and AC components of a candidate block, the DC component having no frequency content and the AC components having frequency content;

selecting a region of the transformed motion compensated difference frame, the selected region including a candidate block and a plurality of neighboring blocks;

comparing, in the frequency domain, at least one component of the transform coefficients of the candidate block within the selected region of the transformed motion compensated difference frame to a threshold value to discriminate between content and noise;

determining whether any of the plurality of neighboring blocks surrounding the candidate block have content or noise; and

encoding the candidate block only if content is detected in the candidate block or in any of the plurality of neighboring blocks.

2. The method of claim 1 , wherein the threshold value is a predetermined static threshold value.

3. The method of claim 1 , wherein comparing the at least one component of the candidate block to the threshold value comprises comparing the DC component of the candidate block to the threshold value.

4. The method of claim 3 , wherein applying the orthogonal transform comprises applying a discrete cosine transform (DCT) to the motion compensated difference frame to obtain the DC component and AC components of the candidate block.

5. The method of claim 4 , wherein comparing the at least one component of the candidate block to the threshold value further comprises comparing at least one AC component of the candidate block to a second threshold value.

6. The method of claim 5 , wherein comparing the at least one AC component of the candidate block to the second threshold value comprises comparing an energy component of an AC region to the second threshold value.

7. The method of claim 1 wherein encoding the candidate block includes encoding to H.263, H.264, or MPEG standards.

8. A content discriminator for a digital video encoder, comprising:

a comparator to compare, in a frequency domain, at least one component of transform coefficients of a candidate block within a selected region of a transformed difference frame to a predetermined threshold value, the selected region including the candidate block and a plurality of neighboring blocks; and

a content detector to discriminate between content and noise based on a result output by the comparator,

wherein the comparator operates to compare at least one component of transform coefficients of the plurality of neighbouring blocks within the difference frame to the predetermined threshold value, and the content detector operates to discriminate between content and noise in the neighbouring blocks.

9. The content discriminator of claim 8 , wherein the predetermined threshold value is a static threshold value.

10. The content discriminator of claim 8 , wherein the at least one component is a DC component of the candidate block.

11. The content discriminator of claim 10 , wherein the DC component is a coefficient of a discrete cosine transform (DCT).

12. The content discriminator of claim 8 , wherein the at least one component is at least one AC component of the candidate block.

13. An video data compressor, comprising:

a transform processor to apply an orthogonal transform to a difference frame of a digital video sequence in a pixel domain to obtain a transformed difference frame having transform coefficients in a frequency domain;

a content discriminator to discriminate between noise and content within a selected region of the transformed difference frame, the selected region including a candidate block and a plurality of neighboring blocks, by comparing, in the frequency domain, at least one component of transform coefficients of the candidate block to a predetermined threshold value, and to determine whether any of the plurality of neighboring blocks have content or noise; and

an encoder to encode the candidate block only if the content discriminator determines content is present in the candidate block or in any of the plurality of neighboring blocks.

14. The video data compressor of claim 13 , wherein the threshold value is a predetermined static threshold value.

15. The video data compressor of claim 13 , wherein the at least one component is a DC component.

16. The video data compressor of claim 15 , the orthogonal transform is a DCT transform.

17. The video data compressor of claim 16 , wherein the at least one component is at least one AC component.

18. The method of claim 6 wherein the energy component of the AC region comprises a sum of squared coefficients of the candidate block.

19. The method of claim 18 wherein the energy component of the AC region comprises a sum of squared values of a subset of AC components of the candidate block.

20. The method of claim 1 wherein the neighbouring blocks comprise eight neighbouring blocks surrounding the candidate block.