Human Body Measurement Extraction from 2D Images

摘要

This work is concerned in extracting the human body measurements with non-contact method. It uses computer vision technique in obtaining the measurements. This method requires background of the input image to be green only. Further segmentation can be generalized so that measurements can be taken with the help of any kind of background. Also, while selecting the alternate to the green segmentation method, it is necessary to consider that it needs the required level of accuracy in segmentation. In the proposed algorithm, key point estimation is done both heuristically and manually. Heuristic key point extraction is of less favorable due to the reason that its incorrect recognition of point might lead to large amount of error in the result. Whereas, on the other hand, manual marking of the key point is also considered of less accuracy. Hence, robust key point estimation method can be implemented rather than the one which is used.