本文通过分析纹理图像经过小波分解后的能量变化特点,比较和分析了利用树型小波变换和塔式小波变换进行图像检索的方法,提出了基于树型小波变换的相似性度量方法。通过排序的评价方法进行评价,研究表明,树型小波分解方法具有较好的纹理区分能力。
This paper compared the pyramid-structured and the tree-structured wavelet transform by analyzing the characterization of energies of image subbands after wavelet transform. Tree,structured wavelet transform (TSWT) and suitable similarity measurement have been developed for retrieval of textures. It proves that effective texture discrimination can be achieved. Experimental results show a superior performance in terms of normalized rank sum.