【摘要】 针对辐射图像的特点,设计并开发了一种支持向量机(SVM)和基于离散小波变换(DWT)的局部特征识别方法。使用支持向量机解决了辐射图像局部特征提取的困难和分类器对样本数目的要求。而小波的应用使得该算法能够支持多分辨率的特征提取,并提高了总体识别效率。还对比了两种常见的核函数,实验结果表明高斯径向基函数能够取得比较好的分类效果。更多还原

【Abstract】 A SVMs (support vector machines) based radiation image local feature recognition algorithm was designed and developed. Using a set of 4000 simulated images, we achieved at least 93. 4% detec tion rate and 0. 8% false positive rate. The empolyment of wavelet, in particluar discrete wavelet trans form (DWT), adds multi-resolution support to the algorithm and incresases total recognition perfor mance. DWT decomposes a radiation image into subbands and stresses features of different scales in each subband. Because of the standout generalization capbility of SVMs, using subbands directly as in put becomes possible and produces compellent results. In this paper, different kernel functions are also compared to each other. Experiments show that Gaussian Raial Basis Function (RBF) kernel overtakes the others in our application.更多还原