Automated Chin-up Testing System Development Combined With Depth Image

Chin-up exercise is a common sport, in its test requires an objective and standard evaluation. Artificial evaluation exists subjective bias and counting, timing error, and large-scale test easily fatigue, low efficiency and fairness is not strong. This paper studies the combination of depth image chin-up project of automatic test technology, the development of the automatic test system based on Microsoft Kinect image system.Chin-up sport was classified as the national high school entry examination sport test items, the greater the demand for its automated testing. The test content is:from the examinee hands are holding bar, the body is still straight arm hanging began to count, statistics of the number of qualified chin-up. A qualified chin-up action required:two arms pull up to the jaw across the bar on the edge, then restore the straight arm suspension. Based on the traditional digital image processing, gesture recognition technology of automatic test, due to the complex background and horizontal bar block face, segmentation of the horizontal bar and jaw position can be difficult. There is a big difference in light intensity in different test environments, but also the traditional image acquisition and processing results prone to error.This paper studies the combination of depth image chin-up automatic test. By using Kinect to obtain depth image, according to the depth information combined with color image for head tracking and image segmentation to identify the position of the horizontal bar and face lower jaw. Before the test, we need to determine the height of the horizontal bar through depth information, for each object at the beginning of the test through the face tracking and head tracking determine its jaw to head distance, thus effectively judge whether jaw level stands above the horizontal bar. Head skeleton tracking results are also used in arm unbend judgment. By the beginning:of the first test, measure the shoulder positioning and horizontal bar maximum height difference, multiplied by the coefficient as a test arms straight threshold, in the test, through head skeletal tracking acquire the jaw position, calculate with the difference in height between the horizontal bar, more than the threshold value is judged to have straight.On the basis of theoretical research, this paper introduces the use of Microsoft’s Kinect image system research chin-up automatic testing. Kinect system can obtain color image and depth image at the same time, the depth of double threshold method is used to remove the complex background,then determine horizontal bar position, thus determine whether chin above the bar; Meanwhile, use the shoulder positioning to determine arm length, comparing the jaw and horizontal bar distance with arm length to determine if exist arms-straight.Development of the system also implements the timing, counting, tweeting, recording and test results saving. Preliminary experiments proved the validity of the auto testing system that it can accurately test chin-up sport project in real time.