Expression Classification of Stroke Patients Based on Improved Edge Detection Mask Algorithm and Neural Network

Stroke is the most common disease of old ages causing serious damage to human life and health. Facial recognition classification is very important in early rehabilitation feedback of stroke patients during training. In expression recognition and classification, the fuzzy expression image and unclear features will lead to the poor quality of stroke patient expression recognition effect. This paper combines image edge detection technology with image mask, proposes an improved edge detection mask algorithm (IEDM). The gradient amplitude is detected by using the pixel domain information of 8 surrounding points and combining global and local feature detection with residual neural network to recognize and classify the facial expression of stroke patients. Through the improved algorithm, it can be concluded that the recognition accuracy of the algorithm used in this paper reaches 90.16%, this algorithm has good edge detection and classification ability.