Study On Pulse Signal Detection Method By Monocular Vision

Pulse is a common physiological phenomenon and the pulse embodies the abundant physiological and pathological information about the health status. The pulse wave is called pulse tracings information in the traditional Chinese medicine (TCM). The subjectivity of pulse tracings' interpretation affects the accuracy and the feasibility of the diagnosis, and that is also the restricted factor in the application and development of pulse diagnosis. Over a long period of time, many researchers devoted to study the different measuring instruments and detecting methods. Nowadays, different kinds of apparatus and sensors were applied to detect the pulse signal such as the liquid pressure sensor, photoelectric sensor, strain sensor and PVDF sensor etc. However, available detecting point obtained by above sensors is limited. Because of the limits in measuring principle, the sensitivity of the elements and the fabrication technology, they can not fulfill the multi-point pulse signals detecting. The objective study of TCM shows that, besides the information of pulse pressure, the pulse has information include pulse tube volume, blood speed and 3D motion of pulse tube etc. So detecting the pulse information by pressure pulse sensors can not reflect the whole information of pulse quantitatively.With the rapidly development of the digital image processing and computer vision technology, image sensor is wildly used in the non-destruction detecting and other fields. In order to obtain the complete and objective pulse tactile information, a pulse image sensor is developed based on monocular vision.In this article, the pulse dynamic image analysis was discussed in detail. At first, the 3-D motion analysis was applied in the pulse motion, the results just show the simply status of pulse motion and can not acquire the pulse wave. Then, the image match technology was studied. The methods of the digital image correlation technology and digital image difference technology were presented. By applying the two methods, the pulse wave can be acquired. Further analysis on the different area reveals the multi-point pulse signal. The signal analysis based on the wavelet analysis suggests that the accumulative difference methods be worthy of researching further reasons. Next, the methods of 'depth and shape from monocular camera' were studied. Furthermore, 'the depth from defocus and the shape from defocus' were presented and applied in the pulse wave and 3-D pulse shape acquirement. The results show that the pulse depth changes and 3-D pulse shape can be acquired by the method.Finally, it is proved through a quantitative experiment that the digital image correlation analysis, digital image difference analysis and defocus analysis for the pulse dynamic image can reflect the out-of-plane displacement quantitatively.As a conclusion, by using the pulse image sensor, the more complete pulse information can be acquired, which is impossible for other pulse sensors. Moreover, it provides a new idea and approach for the objective study on the pulse diagnosis.