Abstract: A high precision sub-pixel spatial alignment of digital images, one from a reference video signal and another from a corresponding test video signal, uses an iterative process and incorporates spatial resampling along with basic correlation and estimation of fractional pixel shift. The corresponding images from the reference and test video signals are captured and a test block is overlaid on them at the same locations to include texture from the images. FFTs are performed within the test block in each image, and the FFTs are cross-correlated to develop a peak value representing a shift position between the images. A curve is fitted to the peak and neighboring values to find the nearest integer pixel shift position. The test block is shifted in the test image by the integer pixel shift position, and the FFT in the test image is repeated and correlated with the FFT from the reference image.

Abstract: A high precision sub-pixel spatial alignment of digital images, one from a reference video signal and another from a corresponding test video signal, uses an iterative process and incorporates spatial resampling along with basic correlation and estimation of fractional pixel shift. The corresponding images from the reference and test video signals are captured and a test block is overlaid on them at the same locations to include texture from the images. FFTs are performed within the test block in each image, and the FFTs are cross-correlated to develop a peak value representing a shift position between the images. A curve is fitted to the peak and neighboring values to find the nearest integer pixel shift position. The test block is shifted in the test image by the integer pixel shift position, and the FFT in the test image is repeated and correlated with the FFT from the reference image.

Abstract: A joint spatial-temporal alignment of video sequences uses a subsampled version of a reference video sequence for correlation with a corresponding subsampled version of a test video sequence in an iterative manner. The subsampled reference video sequence is obtained by detecting a temporal transition field in the reference video sequence, and then selecting a pair of fields from the reference video sequence that straddle the transition field separated by a given number of fields (time interval). Then starting from a current field of the test video sequence a pair of fields is selected separated by the same number of fields and correlated with the subsampled reference video sequence. If the correlation criteria are not satisfied, the current field in the test video sequence is incremented by one and the correlation process repeated until spatial-temporal alignment is achieved.

Abstract: Refinement of motion block motion vectors to achieve a dense motion field for temporal video signal processing makes use of candidate motion vectors from a parent block and neighboring child block groups to determine the motion vector for a child block within one of the child block groups. The vicinity of each candidate motion vector may be searched to obtain a trial motion vector having a minimum displaced frame difference measure, with the trial motion vector with the minimum displaced frame difference measure from among all the trial motion vectors being the motion vector for the child block. Alternatively each candidate motion vector may be applied to the child block, and the vicinity of the candidate motion vector having the minimum displaced frame difference being searched to obtain the motion vector for the child block.

Abstract: Block matching for picture motion estimation uses Gray codes, and preferably a new Hamming-2 Gray code where the bit pattern changes by only one bit for each step in the Gray code count sequence with the additional condition that all other codes are at least two bits different if not adjacent in the count sequence. The pixels for a reference block from a first picture frame and the pixels for a corresponding block within a search area of the next picture frame are each quantized and then converted to an appropriate Gray code. The Gray codes for corresponding pixels are compared to determine whether they are equal or adjacent to each other. If there is adjacency or equality a match signal is generated for that pixel which is tallied. The total tally for each reference block position within the search area is determined, with the maximum tally indicating a match between blocks from which picture motion estimation may be determined.