Encoding video images using foreground/background segmentation

Claim 2

A computer-implemented process for encoding images, comprising the steps of: providing interframe differences between pixels of an original image and a set of reference pixels and thresholding the interframe differences to generate an initial pixel-level mask for the original image comprising a plurality of pixel-level mask elements, each indicating one of a background and a non-background status for a corresponding pixel of the original image; (b) using a first morphological filter to filter out isolated elements in the initial pixel-level mask to generate a filtered pixel-level mask; (c) thresholding blocks of mask elements of the filtered pixel-level mask to generate an initial block-level mask comprising a plurality of block-level mask elements, each indicating one of the background and the non-background status for a corresponding block of pixels of the original image, said thresholding comprising the step of setting each element of the initial block-level mask to indicate the non-background status only if a number of pixel-level mask elements in a corresponding block of the filtered pixel-level mask exceeds a specified threshold value, and otherwise setting said each element of the initial block-level mask to indicate the background status; (d) using a second morphological filter to filter out isolated elements in the initial block-level mask to generate a filtered block-level mask, and (e) encoding the image based on the filtered block-level mask.

Claim 3

The process of claim 1, wherein step (a) comprises the steps of: (1) selecting a threshold value for each component plane of the original image; (2) thresholding interframe difference for each component plane based on the corresponding selected threshold value; and (3) generating the initial pixel-level mask based on the thresholded component planes.

Claim 4

The process of claim 1, wherein the set of reference pixels comprises a reference block for each block of the original image. where each reference block for each block is generated by averaging corresponding pixels for corresponding blocks from a plurality of previous frames.

Claim 5

The process of claim 3, wherein the first morphological filter is defined by: ##EQU11## wherein: M p is the initial pixel-level mask; M' p is the filtered pixel-level mask; "*" denotes convolution; and ##EQU12##

Claim 6

The process of claim 1, wherein the second morphological filter is defined by: wherein: M b is the initial block-level mask; M' b is the filtered block-level mask; ∪ designates an OR operation: "*" denotes convolution; ##EQU13##

Claim 7

The process of claim 5, wherein: step (a) comprises the steps of: (1) selecting a threshold value for each component plane of the original image; (2) thresholding interframe difference for each component plane based on the corresponding selected threshold value; and (3) generating the initial pixel-level mask based on the thresholded component planes; the first morphological filter is defined by: ##EQU14## wherein: M p is the initial pixel-level mask; M' p is the filtered pixel-level mask; "*" denotes convolution; and ##EQU15##

Claim 10

The apparatus of claim 7, wherein means (a): (1) selects a threshold value for each component plane of the original image; (2) thresholds interframe difference for each component plane based on the corresponding selected threshold value; and (3) generates the initial pixel-level mask based on the thresholded component planes.

Claim 11

The apparatus of claim 7, wherein the set of reference pixels comprises a reference block for each block of the original image where each reference block for each block is generated by averaging corresponding pixels for corresponding blocks from a plurality of previous frames.

Claim 12

The apparatus of claim 9, wherein the first morphological filter is defined by: ##EQU16## wherein: M b is the initial pixel-level mask; M' p is the filtered pixel-level mask; "*" denotes convolution; and ##EQU17##

Claim 13

The apparatus of claim 7, wherein the second morphological filter is defined by: wherein: M b is the initial block-level mask; M' b is the filtered block-level mask; ∪ designates an OR operation; "*" denotes convolution; ##EQU18##

Claim 14

The apparatus of claim 11, wherein: means (a): (1) selects a threshold value for each component plane of the original image; (2) thresholds interframe difference for each component plane based on the corresponding selected threshold value; and (3) generates the initial pixel-level mask based on the thresholded component planes; the first morphological filter is defined by: ##EQU19## wherein: M p is the initial pixel-level mask; M' p is the filtered pixel-level mask; "*" denotes convolution; and ##EQU20##

Claim 15

A storage medium having stored thereon a plurality of instructions for encoding images, wherein the plurality of instructions, when executed by a processor, cause the processor to perform the steps of: (a) providing interframe differences between pixels of an original image and a set of reference pixels and thresholding the interframe differences to generate an initial pixel-level mask for the original image comprising a plurality of pixel-level mask elements each indicating one of a background and a non-background status for a corresponding pixel of the original image; (b) using a first morphological filter to filter out isolated elements in the initial pixel-level mask to generate a filtered pixel-level mask; (c) thresholding blocks of mask elements of the filtered pixel-level mask to generate an initial block-level mask comprising a plurality of block-level mask elements, each indicating one of the background and the non-background status for a corresponding block of pixels of the original image said thresholding comprising the step of setting each element of the initial block-level mask to indicate the non-background status only if a number of pixel-level mask elements in a corresponding block of the filtered pixel-level mask exceeds a specified threshold value, and otherwise setting said each element of the initial block-level mask to indicate the background status; (d) using a second morphological filter to filter out isolated elements in the initial block-level mask to generate a filtered block-level mask; and (e) encoding the image based on the filtered block-level mask.

Claim 16

A storage medium having stored thereon a plurality of instructions for encoding images, wherein the plurality of instructions, when executed by a processor, cause the processor to perform the steps of: (a) providing interframe differences between pixels of an original image and a set of reference pixels and thresholding the interframe differences to generate an initial pixel-level mask for the original image comprising a plurality of pixel-level mask elements each indicating one of a background and a non-background status for a corresponding pixel of the original image; (b) using a first morphological filter to filter out isolated elements in the initial pixel-level mask to generate a filtered pixel-level mask; (c) thresholding blocks of mask elements of the filtered pixel-level mask to generate an initial block-level mask comprising a plurality of block-level mask elements, each indicating one of the background and the non-background status for a corresponding block of pixels of the original image said thresholding comprising the step of setting each element of the initial block-level mask to indicate the non-background status only if a number of pixel-level mask elements in a corresponding block of the filtered pixel-level mask exceeds a specified threshold value, and otherwise setting said each element of the initial block-level mask to indicate the background status; (d) using a second morphological filter to filter out isolated elements in the initial block-level mask to generate a filtered block-level mask; and (e) encoding the image based on the filtered block-level mask.

Claim 17

The storage medium of claim 13, wherein step (a) comprises the steps of: (1) selecting a threshold value for each component plane of the original image; (2) thresholds interframe difference for each component plane based on the corresponding selected threshold value; and (3) generating the initial pixel-level mask based on the thresholded component planes.

Claim 18

The storage medium of claim 13, wherein the set of reference pixels comprises a reference block for each block of the original image, where each reference block for each block is generated by averaging corresponding pixels for corresponding blocks from a plurality of previous frames.

Claim 19

The storage medium of claim 15, wherein the first morphological filter is defined by: ##EQU21## wherein: M p is the initial pixel-level mask; M' p is the filtered pixel-level mask; "*" denotes convolution; and ##EQU22##

Claim 20

The storage medium of claim 13, wherein the second morphological filter is defined by: wherein: M b is the initial block-level mask; M' b is the filtered block-level mask; ∪ designates an OR operation; "*" denotes convolution; ##EQU23##

Claim 21

The storage medium of claim 17, wherein: step (a) comprises the steps of: (1) selecting a threshold value for each component plane of the original image; (2) thresholding interframe difference for each component plane based on the corresponding selected threshold value; and (3) generating the initial pixel-level mask based on the thresholded component planes; the first morphological is defined by: ##EQU24## wherein: M p is the initial pixel-level mask; M' p is the filtered pixel-level mask; "*" denotes convolution; and ##EQU25##