Back to results
US 6,636,565 B1Granted
Method for concealing error
AssigneeLG Electronics Inc.
InventorEung Tae Kim
Life of this patent
Estimating the term…
Abstract
A moving picture compression/decompression system utilizes a method for concealing errors in a received picture. A macro block, lost due to an error, is compensated using either a temporal or spatial prediction method, depending upon the type of frame and the characteristics of the picture. Thus, the displayed picture quality can be significantly improved.
Claims9
A method for concealing errors comprising:
- (a) extracting a picture coding type from among I-picture, P-picture and B-picture coding types of an input frame in which an error is generated by searching for coding modes of available neighboring blocks in a current frame to determine characteristics of a bitstream currently being decoded, and obtaining picture characteristics of the input frame, wherein said obtaining the picture characteristics by the following equation where Ct(k) for {k=1, . . . , m} represents neighboring blocks of a block with an error in the input frame at time t and m indicates a number of available neighboring blocks;
- DF = ∑ k = 1 m [LeftBracketingBar] C t ( k ) - C t - 1 ( k ) [RightBracketingBar]
- (b) selecting an error concealing method from one of a plurality of error concealing methods depending upon the picture coding type extracted from among I-picture, P-picture and B-picture coding types and the picture characteristics wherein a temporal prediction error concealing method is selected if the value of DF is less than a first threshold value;
- and
- (c) compensating the input frame by concealing the error using a selected error concealing method.
2 more elements
A method for concealing errors comprising:
- (a) extracting a picture coding type from among I-picture, P-picture and B-picture coding types of an input frame in which an error is generated by searching for coding modes of available neighboring blocks in a current frame to determine characteristics of a bit stream currently being decoded, and obtaining picture characteristics of the input frame;
- (b) selecting an error concealing method from one of a plurality of error concealing methods depending upon the picture coding type extracted from among I-picture, P-picture and B-picture coding types and the picture characteristics, wherein a temporal prediction error concealing method is selected when the frame has a strong temporal correlation, and a spatial prediction error concealing method is selected when the frame has a weak temporal correlation, said spatial prediction error concealing method using a frequency domain interpolation method, wherein the frequency domain interpolation method comprises:
- dividing a block with an error in the input frame into first subblocks and dividing each neighboring block of the block with an error into second subblocks;
- DCT-converting, for each of the first subblocks, the second subblocks neighboring a corresponding one of a first subblock;
- and
- inverse-DCT-converting, for each DCT-converted second subblocks, an average value of coefficients in a low frequency band to obtain pixel values of a corresponding first subblock;
- and
- (c) compensating the input frame by concealing the error using a selected error concealing method.
5 more elements
A method for concealing errors comprising:
- (a) extracting a picture coding type from among I-picture, P-picture and B-picture coding types of an input frame in which an error is generated by searching for coding modes of available neighboring blocks in a current frame to determine characteristics of a bitstream currently being decoded, and obtaining picture characteristics of the input frame;
- (b) selecting an error concealing method from one of a plurality of error concealing methods depending upon the picture coding type and the picture characteristics wherein a spatial prediction error concealing method is selected when the frame has a weak temporal correlation, and a temporal prediction error concealing method is selected when the frame has a strong temporal correlation, said temporal prediction error concealing method using a new cost function to predict a temporal correlation between a block with an error in the input frame and neighboring blocks of the block with an error, wherein an optimal motion vector to compensate the block with an error is obtained by a vector which minimizes the following equation, ( V x * , V y * ) = arg min ( V x , V y ) ∈ S ∑ s = 1 4 D ij s ( V x , V y ) + λ E ij s ( V x , V y )
- when D ij 1 ( V x , V y ) = ∑ k = 0 N - 1 ( x 1 ( i + k + V x , j + V y ) - ∑ m = 0 ω a m 1 x ( i + k , j - m ) ) 2 D ij 2 ( V x , V y ) = ∑ k = 0 N - 1 ( x 1 ( i + V x , j + k + V y ) - ∑ m = 0 ω a m 2 x ( i - m , j + k ) ) 2 D ij 3 ( V x , V y ) = ∑ k = 0 N - 1 ( x 1 ( i + k + V x , j + N - 1 + V y ) - ∑ m = 0 ω a m 3 x ( i + k , j + m + N - 1 ) ) 2 D ij 4 ( V x , V y ) = ∑ k = 0 N - 1 ( x 1 ( i + N - 1 + V x , j + k + V y ) - ∑ m = 0 ω a m 4 x ( i + m + N - 1 , j + k ) ) 2 E ij 1 ( V x , V y ) = ∑ k = 0 N - 1 ( x 1 ( i + k + V x , j + V y - 1 ) - x ( i + k , j - 1 ) ) 2 E ij 2 ( V x , V y ) = ∑ k = 0 N - 1 ( x 1 ( i + V x - 1 , j + k + V y ) - x ( i - 1 , j + k ) ) 2 E ij 3 ( V x , V y ) = ∑ k = 0 N - 1 ( x 1 ( i + k + V x , j + N + V y ) - x ( i + k , j + N ) ) 2 E ij 4 ( V x , V y ) = ∑ k = 0 N - 1 ( x 1 ( i + N + V x , j + k + V y ) - x ( i + N , j + k ) ) 2
- wherein x(i,j) represents luminance values of pixels of top, sides and bottom blocks adjacent to a lost block, x1(i,j) represents luminance values of pixels in a previous frame corresponding to x(ij);
- and
- (c) compensating the input frame by concealing the error using a selected error concealing method.
3 more elements
A method for concealing errors comprising:
- (a) extracting a picture coding type from among I-picture, P-picture and B-picture coding types of an input frame in which an error is generated by searching for coding modes of available neighboring blocks in a current frame to determine characteristics of a bit stream currently being decoded, and obtaining picture characteristics of the input frame;
- (b) selecting an error concealing method from one of a plurality of error concealing methods depending upon the picture coding type extracted from among I-picture, P-picture and B-picture coding types and the picture characteristics, wherein the frame is compensated by replacing a block with an error in the frame by a substitute block in which errors are concealed using the selected error concealing method, and a post-process is performed for the substitute block, wherein the post-process includes replacing a boundary pixel of the substitute block by an average value calculated by the following equation, where {overscore (X)}0 represents a pixel value located at the boundary of the substitute block and X1 represents a pixel value adjacent to said boundary pixel at a neighboring block, X _ 0 = X _ 0 + X 1 2 ;
- and
- (c) compensating the input frame by concealing the error using a selected error concealing method.
1 more element
Dependencies are inferred from the claim text.