Author(s)

Zhifeng Wang

Corresponding Author

Zhifeng Wang

Abstract

It is significant for its development and application to launch tai chi image and video detection, and the current image detection is hard to meet the needs. For the accuracy of multi-resolution tai chi image detection, the study put forward a novel estimation method of multi-resolution tai chi image nonparametric kernel density. Firstly, based on the theorem distribution of multi-resolution difference images, kernel density estimation algorithm is used to finish the obtaining of fitting degree optimum value, confirm current test parameters, and differentiate the matching degree of image distribution function in various samples. From a mathematical point of view, that is, we can obtain the optimal value of the wide window of the current difference image in the form of the application of the integral mean-square error, improve the smoothness of kernel density function image curve, reduce the loss of estimation accuracy led by wide window fitting, and get the result of kernel density function by adopting gk kernel function as an initial function for optimal wide window value, thus completing final estimation. After analysis and experiments, it can be determined that, because of adding several optimization programs, the proposed kernel density function evaluation method has higher accuracy and smoother image curve, and it has important application value for capturing tai chi image detection.

Keywords

Image, Kernel density, Standard deviation