基于综合特征的彩色图像检索

Image Retrieval Based on Shape and Local Color Distribution

本文综合利用颜色和形状特征进行基于内容的彩色图像检索.利用边缘方向自相关图表示图像的形状特征.对于颜色特征,计算图像颜色的局部累加直方图,同时提取分块的颜色矩弥补其不包含颜色空间分布关系的缺点.并利用Guassian模型对各特征的距离进行归一化,综合上述三个归一化距离,进行全局相似度量,且用权值调整的相关反馈方法进一步提高检索精度.与局部累加颜色直方图和局部颜色矩的方法相比较,本文提出的方法获得了更好的检索结果.

In this paper, content-based image retrieval using color and shape is studied. To represent the shape content of an image, the edge orientation autocorrelograrn is used. To represent color feature, local color cumulative histogram is computed. We also extract the color moments of partitions to solve the problem of lacking the spatial knowledge. The Guassian model is used to normalize the feature distance respectively. Combining the above three normalized distance, the global similarity measure is obtained, and weight value adjusting improves retrieval precision. We compare the proposed method with local cumulative histogram method and local color moment method . The integrated feature based method of this paper can obtain better retrieval results.