区域加权信息熵及其在图像特征提取中的应用

Regional weighted comentropy and its application in image feature extraction

图像特征提取是当前基于内容图像检索领域的研究重点,然而单纯基于信息熵的图像特征提取方法无法体现图像内容的位置信息。分析现有的基于颜色—空间图像特征提取算法的基础上,结合图像信息熵概念与图像分割算法,提出了一种新的图像信息熵描述方法,即区域加权信息熵,并证明了区域加权信息熵的若干性质。采用信息熵性能评价指标从概率的角度描述因权值变化而引起的图像信息熵分布的变化,并考虑应用的兴趣区域以及权值粒度从而确定合理权值。实验表明区域加权信息熵方法比单纯信息熵方法描述图像内容准确率提高了50%以上。

Image feature extraction is the research focus in the field of content-based image retrieval; however, entropy- based image feature extraction cannot demonstrate the location of image content information. A new description method of image comentropy named regional weighted comentropy was proposed, which combined the concept of image comentropy and image segmentation algorithm after analyzing the current color-space image feature extraction algorithms. Some properties of regional weighted comentropy were proved. The distribution change of image comentropy, which was caused by weight＇s change, was described by using comentropy performance evaluation index in terms of probability, considering the interested regions and weights precision applied by users, then the reasonable weight was determined. Experimental results show that the accuracy of image content described by regional weighted comentropy method is more than 50% higher than that of traditional comentropy methods.