基于Radon和小波变换的图像检索

IMAGE RETRIEVAL BASED ON RADON AND WAVELET TRANSFORM

提出一种基于Radon和小波变换的图像纹理特征检索算法.对原始图像进行坐标系的方向归一化,再对方向归一化后的图像进行Radon变换.根据Radon变换投影数据的几何特性,构造了适合投影数据的具有尺度和平移不变性的小波分解,该小波分解系数具有旋转、平移和尺度不变性.采用图像中各尺度小波系数的能量值作为图像的纹理特征,以此作为纹理特征进行图像检索.基于纹理特征的试验结果表明该特征具有旋转、平移和尺度不变性,与其他算法相比具有较高的检索率.

A new kind of image retrieval algorithm based on the texture feature was presented by using the Radon and wavelet transform. Firstly, the orientation-normalized image was achieved by rotating its principal axis to horizontal direction and the Radon transform was used to deal with the orientation-normalized image. It can get the rotation-invariant Radon project of the image. Secondly, the new adoptive wavelet decomposition for the Radon project of the image was constructed according to the property of Radon transform, which had the property of shift- and scale-invariant to the project of Radon transform. Then the wavelet coefficients had the property of rotation-, translation- and scale-invariant to the project of Radon transform. Finally, the energy of every sub frequency bands was extracted to be the texture feature of image. The image retrieval experiments indicate that this method has the property of shift, scale and rotation invariance and it has a higher retrieval-rate than other methods＇.