摘要
利用图像分形编码中定义域块和最小均方误差一一对应的特点 ,提出了 1种基于分形编码的多姿态、表情的人脸图像检索方法。该方法将待检索图像分割为相同大小的值域块 ,然后将每一值域块按给定的定义域块进行分形编码 ,得到最小均方误差 ,计算该最小均方误差与图像库中最小均方误差的欧氏距离。将待检索图像所有值域块的欧氏距离求平均 ,此平均欧氏距离较小的几幅图像即为检索出的图像。实验证明该方法能够准确地检索出图像库中存储的同一人的不同姿态、表情的图像。
According to the feature that the domain block is in one-to-one correspondence with the least mean square error (LMSE) in fractal coding, a multi-pose-and-expression face retrieval method is proposed in this paper. The retrieval image is divided into range blocks whose sizes are the same as those stored in the database, then the range blocks are coded in terms of preset domain blocks, which results in an LMSE. Compute the Euclid distance between the above LMSE and the one stored in the database. Average the Euclid distances of all the range blocks in the retrieval image, then the images whose average Euclid distances are smaller are selected as the retrieval images. The experiments prove that the images stored in the database can be retrieved rapidly and exactly.
出处
《中国海洋大学学报（自然科学版）》
CAS
CSCD
北大核心
2004年第5期826-830,共5页
Periodical of Ocean University of China
基金
国家自然科学基金资助项目 (69972 0 1 9)
山东省自然科学基金资助项目 (Y2 0 0 0G0 3)资助
