基于多小波分布熵的图像检索

为了提高小波直方图的检索性能,提出基于多小波信息分布熵的图像检索算法。对检索图像进行多小波分解,并用滤波器对各个子图进行非线性滤波,计算多小波各子带的能量熵;对各子图的小波能量矩阵进行0/1量化,然后以各子带相同方向子图计算多小波分布熵;针对特征向量进行高斯归一化,利用欧氏距离计算不同图像间的纹理相似度。基于内容的图像检索试验表明,该方法的检索精度比快速小波直方图方法提高了9.7%。

Image retrieval based on the multiwavelet entropy

In order to enhance the retrival performance of the wavelet histogram,a new retrieval algorithm based on the multiwavelet entropy is presented.Firstly,the retrieval image is decomposed by the multiwavelet transform,and the nonlinear filter in the wavelet domain is used to deal with the multiwavelet sub-bands.The energy entropy of every sub-band is computed.Secondly,the energy matrices of each sub image of multiwavelet decomposition are 0/1 quantized,then the multiwavelet coefficient entropies are computed as the texture features of images according to the same direction sub images of each sub-band.The Gaussian model is used to normalize the sub-characters of different images.Finally,the texture similarity between the querying image and other images is computed by the Euclidean distance.Experiments indicate that the retrieval rate of the proposed method has raised by 9.7% as compared with that of FWHT.