Recursive Gabor transform for images of very large size and designof massively parallel architecture

摘要

The conventional Gabor transforms, complete and generalized,cannot be applied for image matrices received as different partitions.In practice, very large images are received in the form of submatrixpartitions. Then, taking transform on different partitions (of the sameimage) at different times is of prime importance. The situation can beeasily handled if the transformation algorithm is made recursive. Theauthors present a recursive Gabor transform (RGT) amenable forpipelining partitions of different images interleaved. Apriori knowledgeof image size is not required for the RGT unlike the conventional GaborTransform. A dedicated massively parallel architecture based on PacubeVLSI arrays is presented which achieves a high degree of pipelining withrespect to different image matrices. The pipelining rate is proportionalto the delay of a Pacube macrocell in mode 1 operation