基于面积直方图及交互信息测度论的医学图像检索方法

颜色直方图是基于内容的医学图像检索(CBMIR)中提取图像特征的一种重要方法,然而其完全丢弃了医学图像颜色的空间分布信息。为了有效地利用图像颜色的空间分布信息,本文提出了一种基于面积直方图及交互信息测度法的医学图像检索方法。该方法进一步分析了具有相同灰度的像素之间在空间分布上的差异,用其在图像中形成的主要连通区域的面积大小代替传统直方图中使用的像素总和,从而能够间接反映色彩的空间分布特征;同时,为了提高检索准确率,本文引进了交互信息测度方法。实验结果表明,本文提出的检索方法能够较为准确的反映医学图像的信息特征,从而提高了检索的鲁棒性、准确性和高效性。

The Medical Image Retrieval Based on the Area Histogram and Mutual Information

Color histogram is an important method to extract the feature of the image in content-based medical image retrieval. The conventional color histogram retrieval method is prone to lose the spatial information of colors.For the above feature, a new retrieval algorithm based on the area histogram and mutual information is proposed in this paper for the medical image retrieval. This algorithm analyses the different spatial distribution within the pixels with same gray level and substitute total amount of pixels used in conventional histogram with the prominent region's area value formed by theses pixels to capture the spatial information. At the same time the mutual information is used to improve the precision ratio. The experimental results indicate that the feature of the medical image can be reflected exactly by using the proposed algorithm, which has a good robustness, precision and high efficiency.