DFT-based fast algorithms for 2-D discrete Gabor transform

The 2-D Gabor transform has been recognized as being very useful in diverse areas such as image compression, texture analysis, image segmentation, and image recognition; however, its real-time applications have been limited due to the high computational complexity of the 2-D discrete Gabor transform (DGT) algorithms. The contributions of this paper include: (1) the oversampled 2-D DGT is presented, which can utilize the fast DFT and inverse DFT algorithms for fast computation of the 2-D DGT coefficients of an image and for the fast reconstruction of the original image from the coefficients, and (2) a fast computation method of the biorthogonal analysis windows based on the 2-D biorthogonality relationship of the 2-D DGT is provided. The proposed algorithms are attractive for real time image processing.