This paper proposes a tree-structured wavelet transform for texture analysis and classification. The method is based on the observation that many natural textures can be modeled as quasi-periodic signals with dominant frequencies in the middle frequency channels. The tree-structured wavelet transform allows for decomposition into any desired frequency channels, unlike the conventional pyramid-structured wavelet transform, which only decomposes in low frequency channels. The proposed method is computationally efficient and performs well in texture classification. It is compared with other methods such as DCT, DST, DHT, pyramid-structured wavelet transforms, Gabor filters, and Laws filters. The tree-structured wavelet transform is shown to be more effective for textures with dominant middle frequency channels. The method uses a progressive classification algorithm that starts with one feature and adds more features as needed. Experimental results show that the tree-structured wavelet transform provides better classification performance than other methods, especially for textures with visually similar patterns. The method is also shown to be robust to noisy data and effective for textures that are visually similar. The paper concludes that the tree-structured wavelet transform is a promising tool for texture analysis and classification.This paper proposes a tree-structured wavelet transform for texture analysis and classification. The method is based on the observation that many natural textures can be modeled as quasi-periodic signals with dominant frequencies in the middle frequency channels. The tree-structured wavelet transform allows for decomposition into any desired frequency channels, unlike the conventional pyramid-structured wavelet transform, which only decomposes in low frequency channels. The proposed method is computationally efficient and performs well in texture classification. It is compared with other methods such as DCT, DST, DHT, pyramid-structured wavelet transforms, Gabor filters, and Laws filters. The tree-structured wavelet transform is shown to be more effective for textures with dominant middle frequency channels. The method uses a progressive classification algorithm that starts with one feature and adds more features as needed. Experimental results show that the tree-structured wavelet transform provides better classification performance than other methods, especially for textures with visually similar patterns. The method is also shown to be robust to noisy data and effective for textures that are visually similar. The paper concludes that the tree-structured wavelet transform is a promising tool for texture analysis and classification.