Determination of Anatomy Joint Position in Human Body Combining Human Contour Image Feature Algorithm

Abstract

Abstract How to extract the feature information related to human body structure from image sequences and complete human motion analysis including human posture recognition is highly important research work. In this paper, an algorithm for determining the position of human joints based on human contour image features is proposed. Firstly, the virtual skeleton of the human body is extracted from the human contour using the energy function. Subsequently, the position of joints is determined based on three rules to identify whether a point in the human virtual skeleton is a joint point given in the standard human skeleton model as well as the related knowledge of human anatomy. The experimental results show that the proposed algorithm has no restrictive conditions attached to the human body in the image in the aspects of motion and color, etc. At the same time, it also has an excellent suppression effect on the noise in the human contour image. Attributing to this feature, it has a relatively low requirement for the extraction accuracy of human contour, with relatively good performance in complex backgrounds.