摘要
为了更好地进行图像纹理分析,提出了一种基于Radon变换的不变量纹理识别算法。该算法首先利用Radon变换将图像投影到1维空间,然后通过对投影数据进行一种平移和比例不变的自适应小波变换来构造出具有比例和平移不变性的图像的特征矩阵。这种通过对特征矩阵进行多尺度分析得到的多尺度能量特征不但具有平移、比例和旋转不变性,而且反映出了纹理图像在不同尺度上的能量分布特征。在特征提取完成以后,即可利用支撑向量机进行分类。同其他方法的比较说明,该算法可较好地描述纹理特征,并可完成纹理识别。
For image texture analysis, an invariant texture recognition algorithm is proposed based on Radon transform. Firstly, Radon transform is used to project the image to 1-D space, and then the projection data is transformed via a translation and scaling invariant adaptive 1-D wavelet transform, thus the feature matrix with translation and scaling invariance is derived. Multiscale analysis is employed for the feature matrix, and the energy values at different scales are proven not only to be invariant under image translation, scaling and rotation, but also to reflect the different energy distribution of the texture image at different scales. After the feature vector is availabe, support vector machine ( SVM ) is used for classification. Comparing with other methods simulations are given to obtain insight into the effectiveness of our method.
出处
《中国图象图形学报》
CSCD
北大核心
2005年第9期1117-1123,共7页
Journal of Image and Graphics
基金
国家自然科学基金项目(60372085)
航空科学基金项目(03D53032)
关键词
RADON变换
不变量
多尺度
支撑向量机
radon transform, invariant, multiscale, support vector machine(SVM)