一种基于Radon变换和双谱分析的纹理旋转不变性分析方法  

Rotation Invariant Analysis Approach to Texture Image Classification Based on Radon Transform and Bispectrum Analysis

作　　者：王晅[1] 赵杰[1] 

机构地区：[1]陕西师范大学物理学与信息技术学院,陕西西安710062

出　　处：《铁道学报》2009年第1期103-106,共4页Journal of the China Railway Society

基　　金：国家863计划资助项目(2002AA143021)

摘　　要：现有纹理旋转不变分析方法主要基于纹理图像局部特征,而且对噪声较为敏感。本文提出一种基于Radon变换和双谱分析的纹理旋转不变分析方法。该方法对纹理图像进行Radon变换,在Radon投影空间进行双谱分析实现纹理旋转不变识别与旋转角度估计。由于Radon变换是对图像不同方向的线积分,所以基于Radon投影空间的纹理特征具有全局特性,而且对白噪声具有一定的抑制作用。均值为零的高斯过程其双谱为零,所以此方法对高斯噪声也有抑制作用。本方法的特征向量长度很短,降低了分类器训练和识别的复杂度。实验采用Brodatz纹理集,结果表明:该方法具有较高的识别率,能够较为精确地计算纹理的旋转角度,而且对噪声的鲁棒性强。The existing rotation invariant analysis methods for texture image classification are based on local features of texture images and are sensitive to noise. This paper proposes a new rotation invariant analysis approach to texture image classification. In the proposed approach, the Radon transform is employed to project the texture image to be analyzed onto the projection space, and then bispectrum analysis approach is applied to implement texture rotation invariant classification and calculation of the angle which the texture image is rotated by. The texture patterns extracted from the Radon projection spaces of texture images are global and less sensitive to white noise due to the fact that the Radon transform of images is linear integrals over a family of straight lines for different orientations. Since the bispectrum results of Gaussian noises with zero mean are all zero, the proposed approach is less sensitive to Gaussian noise. Due to the short feature vectors, training and recognition of the classifier become simple. The performance of the proposed approach is evaluated using a texture set from Brodatz album. Experimental results show the high classification accuracy of this method. It is also shown that this proposed approach is relatively robust in the presence of noise.

关 键 词：RADON变换 旋转不变 双谱分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]