摘要： 将单幅图像数据进行分割，获得高维化后的数据集合，再依据图像数据的最优分解来提取不同图像块之间的数字关联，利用多维尺度分析(MDS)方法来获取单幅图像数据不同块之间的低维表示。通过对此低维表示的自动分析，便可获得图像感兴趣区域的精确位置。的实例验证了方法的可行性、有效性。
关键词:
图像感兴趣区域,
高维化,
最优分解,
多维尺度分析
Abstract: A new method on the region-of-interest image is presented, the image is parted. By the greatest expression of the partition, the relations of the different partitions are gained. At last, according to the MDS, the low-dimension is achieved. Through analyzing the low-dimension data, the precision location of the region-of-interest can be denoted. The examples validate the feasibility and validity.
Key words:
Region-of-interest,
High-dimension,
Greatest expression,
Multidimensional scaling
刘红霞;谭 璐;吴 翊. 基于MSD的图像感兴趣区域自动提取方法研究[J]. 计算机工程, 2006, 32(24): 195-197.
LIU Hongxia; TAN Lu; WU Yi. Research on Region-of-interest Image Auto-selection Based on MSD[J]. Computer Engineering, 2006, 32(24): 195-197.