|
|
|
“Texture Analysis and Classification with Tree-Structured Wavelet Transform”
by Tianhorng Chang and C.-C. Jay Kuo
February 1992
Although textures have been studied for more than thirty years, research on texture analysis is still active. The main difficulty of the problem is due to the lack of an adequate tool which characterizes different scales of textures effectively. Traditional methods based on the second-order statistics or the Gaussian Markov Random Field (GMRF) model share one common weakness. That is, they primarily focus on the coupling of pixels in a single scale. Recent developments in multiresolution analysis such as the Gabor and wavelet transforms help to overcome this difficulty. In this research, we propose a multiresolution approach based on a tree-structured wavelet transform for texture classification. The development of tree-structured wavelet transform is motivated by the observation that textures are quasi-periodic signals whose dominant frequencies are located in the middle frequency channels. With the transform, we are able to zoom into desired frequency channels and perform further decomposition. In contrast, the conventional wavelet transform only decomposes subsignals in low frequency channels. We also develop a progressive texture classification algorithm which is not only computationally attractive but also has excellent performance.