Methods, systems and apparatus to improve image classification with boundary-bitmaps

1. An apparatus to analyze an image, the apparatus comprising:

a silhouette engine to:

identify a foreground silhouette within the image;

generate a bounding box based on borders of the foreground silhouette; and

generate an encoded silhouette matrix which identifies cells of a foreground and cells of a background;

a convolution cell selector to convolve the encoded silhouette matrix to generate a convoluted bitmap matrix; and

a filter cell selector to improve image classification efficiency by identifying eligible blocks of the convoluted bitmap matrix by:

retaining first respective cells of the convoluted bitmap matrix that satisfy a cell retention threshold;

removing second respective cells of the convoluted bitmap matrix that do not satisfy the cell retention threshold; and

a classification interface to provide the retained first respective cells of the convoluted bitmap matrix to a histogram of oriented gradients classifier.

2. The apparatus as defined in claim 1 , wherein the filter cell selector is to generate a contributed blocks matrix by encoding the first respective cells of the contributed blocks matrix with a retention indicator, and encoding the second respective cells of the contributed blocks matrix with a discard indicator.

3. The apparatus as defined in claim 2 , further including a boundary encapsulator to identify an outer perimeter of respective ones of the retention indicator to generate an eligible block matrix.

4. The apparatus as defined in claim 3 , further including a classification interface to mask a detection window with the eligible block matrix, the mask to reduce a number of blocks to be processed in image classification.

5. The apparatus as defined in claim 1 , further including a window engine to:

convolve a ones matrix kernel to respective cells of the encoded silhouette; and

calculate a sum total of foreground cells within the ones matrix kernel for the respective cells of the encoded silhouette.

6. The apparatus as defined in claim 5 , wherein the convolution cell selector is to encode the convoluted bitmap matrix with the sum total in respective cells of the convoluted bitmap matrix.

7. The apparatus as defined in claim 1 , wherein the cell retention threshold includes a value based on at least one of a matrix kernel size or an overlap step size.

8. A method to analyze an image, the method comprising:

identifying, by executing an instruction with a processor, a foreground silhouette within the image;

generating, by executing an instruction with the processor, a bounding box based on borders of the foreground silhouette;

generating, by executing an instruction with the processor, an encoded silhouette matrix to identify cells of a foreground and cells of a background;

convolving, by executing an instruction with the processor, the encoded silhouette matrix to generate a convoluted bitmap matrix;

improving image classification efficiency by identifying, by executing an instruction with the processor, eligible blocks of the convoluted bitmap matrix by (a) retaining first respective cells of the convoluted bitmap matrix that satisfy a cell retention threshold, and (b) remove second respective cells of the convoluted bitmap matrix that do not satisfy the cell retention threshold; and

transmitting the retained first respective cells of the convoluted bitmap matrix to a histogram of oriented gradients classifier.

9. The method as defined in claim 8 , further including generating a contributed blocks matrix by encoding the first respective cells of the contributed blocks matrix with a retention indicator, and encoding the second respective cells of the contributed blocks matrix with a discard indicator.

10. The method as defined in claim 9 , further including identifying an outer perimeter of respective ones of the retention indicator to generate an eligible block matrix.

11. The method as defined in claim 10 , further including masking a detection window with the eligible block matrix, the mask to reduce a number of blocks to be processed in image classification.

12. The method as defined in claim 8 , further including:

convolving a ones matrix kernel to respective cells of the encoded silhouette; and

calculating a sum total of foreground cells within the ones matrix kernel for the respective cells of the encoded silhouette.

13. The method as defined in claim 12 , further including encoding the convoluted bitmap matrix with the sum total in respective cells of the convoluted bitmap matrix.

14. The method as defined in claim 8 , further including setting a value of the cell retention threshold based on at least one of a matrix kernel size or an overlap step size.

15. A tangible computer-readable medium comprising instructions that, when executed, cause a processor to, at least:

identify a foreground silhouette within an image;

generate a bounding box based on borders of the foreground silhouette;

generate an encoded silhouette matrix to identify cells of a foreground and cells of a background;

convolve the encoded silhouette matrix to generate a convoluted bitmap matrix;

improve image classification efficiency by identifying eligible blocks of the convoluted bitmap matrix by (a) retaining first respective cells of the convoluted bitmap matrix that satisfy a cell retention threshold, and (b) remove second respective cells of the convoluted bitmap matrix that do not satisfy the cell retention threshold; and

forward the retained first respective cells of the convoluted bitmap matrix to a histogram of oriented gradients classifier.

16. The computer-readable medium as defined in claim 15 , wherein the instructions, when executed, further cause the processor to generate a contributed blocks matrix by encoding the first respective cells of the contributed blocks matrix with a retention indicator, and encoding the second respective cells of the contributed blocks matrix with a discard indicator.

17. The computer-readable medium as defined in claim 16 , wherein the instructions, when executed, further cause the processor to identify an outer perimeter of respective ones of the retention indicator to generate an eligible block matrix.

18. The computer-readable medium as defined in claim 17 , wherein the instructions, when executed, further cause the processor to mask a detection window with the eligible block matrix, the mask to reduce a number of blocks to be processed in image classification.

19. The computer-readable medium as defined in claim 15 , wherein the instructions, when executed, further cause the processor to:

convolve a ones matrix kernel to respective cells of the encoded silhouette; and

calculate a sum total of foreground cells within the ones matrix kernel for the respective cells of the encoded silhouette.

20. The computer-readable medium as defined in claim 19 , wherein the instructions, when executed, further cause the processor to encode the convoluted bitmap matrix with the sum total in respective cells of the convoluted bitmap matrix.

21. The computer-readable medium as defined in claim 15 , wherein the instructions, when executed, further cause the processor to set a value of the cell retention threshold based on at least one of a matrix kernel size or an overlap step size.

22. A system to analyze an image, the system comprising:

means for identifying a foreground silhouette within the image;

means for generating a bounding box based on borders of the foreground silhouette;

means for generating an encoded silhouette matrix to identify cells of a foreground and cells of a background;

means for convolving the encoded silhouette matrix to generate a convoluted bitmap matrix;

means for improving image classification efficiency by identifying eligible blocks of the convoluted bitmap matrix by (a) retaining first respective cells of the convoluted bitmap matrix that satisfy a cell retention threshold, and (b) remove second respective cells of the convoluted bitmap matrix that do not satisfy the cell retention threshold; and

means for classification to provide the retained first respective cells of the convoluted bitmap matrix to a histogram of oriented gradients classifier.

23. The system as defined in claim 22 , further including means for generating a contributed blocks matrix by encoding the first respective cells of the contributed blocks matrix with a retention indicator, and encoding the second respective cells of the contributed blocks matrix with a discard indicator.

24. The system as defined in claim 23 , further including means for identifying an outer perimeter of respective ones of the retention indicator to generate an eligible block matrix.

25. The system as defined in claim 24 , further including means for masking a detection window with the eligible block matrix, the mask to reduce a number of blocks to be processed in image classification.

26. The system as defined in claim 22 , further including means for:

convolving a ones matrix kernel to respective cells of the encoded silhouette; and

calculating a sum total of foreground cells within the ones matrix kernel for the respective cells of the encoded silhouette.