sEMG activities detection by improved Sobel algorithm

How to auto-complete surface electromyography (sEMG) activities detection, and improve its accuracy is an important prerequisite to achieve real-time and effective control of myoelectric prosthesis. In this paper, activities detection problem can be equivalent to the edge detection problem in image processing. Taking the advantage of the edge maximum a posteriori estimates, a threshold set is proposed to improve the Sobel operator. Meanwhile, according to the certain similarity between the activities detection and speech endpoint detection, the improved Roberts edge detection, and two kinds of pattern recognition methods - automatic clustering and minimum error rate Bayes classification which were used in speech processing are applied to try sEMG activities detection. The comparison experiments of four methods are carried out. The experimental results show that the four algorithms can achieve the sEMG activities detection automatically, and in which the improved Sobel algorithm has the best test results.