The Algorithm for Extracting Elements of National Costume Based on Region Growing

Deqiang Yang, Jianhou Gan, Bin Wen, Tianwei Xu
2017 Proceedings of the 2017 2nd International Conference on Control, Automation and Artificial Intelligence (CAAI 2017)   unpublished
It is difficult to effectively extract costume elements from the national costume image because the structure of costume images is complex and the gray distribution is irregular. Therefore, in this paper, a novel algorithm was proposed for the extraction of the elements using the mathematical morphology method for element internal filling and external pruning. An improved algorithm for region growing image segmentation was proposed to dynamically calculate the mean and variance of seed regions
more » ... s growth parameters. According to the elements of the gray dispersion, the seeds were divided into three kinds of them to enter the target image which need to be extracted. Different types of seeds played different roles in the growth process. The target element were gradually obtained with each seed growth. The experimental results show that the relatively complete and accurate elements of the costume can be extracted by the algorithm.
doi:10.2991/caai-17.2017.71 fatcat:xldk2i4a2bapnbybh3jpiobbpa