基于树状小波变换后分区加权的图像检索方法

An Image Retrievel Method Based on Partition and Weighting after Tree-Structured Wavelet Transform

论文提出了一种基于树状小波变换后分区加权的彩色图像检索方法。运用树状小波变换对图像进行分解,将图像分解成一系列的高频图像和低频子图。高频图像反映了图像的纹理细节。在计算纹理特征时,对高频子图进行了分区加权,突出了图像中间部分的纹理特征;采用HSV颜色空间的直方图来表示图像的颜色特征。综合检索时,对这两种特征进行归一化处理。实验结果表明,这种方法是有效的。

In this paper we put forward a new method for image retrievel which is based on partition and weighting after Tree-Structured Wavelet Transform.we use Tree-structured Wavelet Transform decomposing the image into some high-frequency image and some low-frequency image.High-frequency image can reflect the texture of the image,so statisticsing the high-frequency image,we can get the texture feature.When computing the texture feature,we divide the high-frequency image into three parts,highlighting the middle part of the image.We use the histogram of the HSV color histogram as the color feature.During retrievel,we standardize two kind of feature.Experimental results show that this method is efficient.