基于人体分割的智能女装廓形尺寸数据库构建

为准确地对现有秀场女装廓形进行判断,提出一种基于人体分割的智能女装廓形尺寸数据库的构建方法和流程。针对静态图像,先进行人脸检测,再利用最大后验概率模型检测人体,并通过图割技术提取服装廓形,最终确定人体影响服装廓形的关键部位,自动测量服装廓形相关尺寸并入库。针对各种可能遇到的实际图像情况提出相应的测量准则,并提炼出显著影响服装廓形的“突变点”概念。经验证,该方法能够有效快速地对静态图像中的女装廓形批量提取廓形尺寸数据,测量数据共包括8个关键点测量值,具有较高的准确性和实践性。

Smart database of women’s clothing silhouette dimensions based on human body segmentation

Estabishment method andprocess a smart database designed for women’s clothing silhouettes were proposed to accurately judge the existing women's fashion profile. The first step was to detect the human face from a static figure. Then, the human body was characterized by the maximum posteriori probability model. Thirdly, the foreground models from the original figures were extracted with the graph cuts algorithm. The final step includes the definition of the key points of the key points of the generalized silhouettes, the measurement of the corresponding dimensions, and the process to archive the data. Intelligent measurement criterion focused on variousfurther presented special cases. Besides, a meaningjul concept pf "drastic change point" was proposed, whichwas proved important to numerically handle a wider range of women's clothng. This proposed method can be used for effectively measuring 8 majordimensions of women's clothing silhouette from given static images, which could be further applied to estimate the future silhouette trends in the field of fashion design.