Tamura纹理特征在水下目标分类中的应用

Application of Tamura texture feature to classify underwater targets

水下高速目标的功率谱图与非高速目标存在一定的差异性。本文介绍了Tamura纹理特征中粗糙度、对比度和方向度的计算方法。而后分别计算了非高速目标和高速目标的Tamura纹理特征量的三个特征量,通过比较发现其在对比度上存在较为明显的差异性。根据支持向量机分类结果进行了特征量选择分析研究,从分类结果来看,利用粗糙度和对比度两个特征量可以较好的分辨高速目标与非高速目标。

There is some distincmess between underwater high and low speed targets on power-spectra. In this paper, the Coarseness, Contrast and Directionality arithmetic methods of Tamura texture characteristics are introduced. Then, the three Tamura texture characteristics of underwater high and low speed targets are calculated and there is obvious distinctness on contrast. The characteristic choice is investigated according to the classification consequences of Support Vector Machine. It can been seen from the classification consequence that Coarseness and Contrast could well distinguish the high and low speed targets.