A detection model generation method for inner plane rotation of gestures based on hog+svm framework

Abstract

The invention discloses a method for generating a gesture inner plane rotation detection model based on the HOG+SVM framework. The web camera is used to continuously capture the image frame in front of the mirror, and then the detection framework based on the HOG+SVM is used to judge whether the frame contains specified Gestures and their positioning within the frame. The method of the present invention includes: collecting specified gesture data sets in several specific direction ranges, extracting the HOG features of each direction data set and using the SVM method to train the SVM model, generating SVM models in other ranges of direction in the plane space, and adjusting the SVM according to the actual situation Model parameters and input into a multi-model detection application based on the HOG+SVM framework. The invention provides a conversion method of the SVM model of the predetermined rotation direction, and shows that the invention can complete the task of detecting the rotation in the specified gesture plane on the basis of ensuring the detection accuracy through theoretical support and actual experimental results.

Description

technical field [0001] The invention relates to a method for generating a gesture inner plane rotation detection model based on a HOG+SVM framework, and belongs to the technical fields of image processing and pattern recognition. Background technique [0002] With the popularization and rapid development of computers, people's lives are inseparable from the use of computers. As we all know, the previous way of interaction between human and computer is through specific input and output devices, such as keyboard, mouse, stylus, scanner and so on. However, with the development of multimedia technology and the continuous improvement of computer performance in recent years, personal computers (PCs) already have basic functions of various communication media such as voice processing and graphic image processing. In order to improve the ease of use of computers and the naturalness of human-computer interaction, new input technology has increasingly become a research hotspot that h...

AI Technical Summary

Problems solved by technology

The traditional vision-based gesture detection method has the following disadvantages: the background and the human hand area are required to have a good distinction in color space, and environmental factors such as the lighting environment cannot cause too much interference to the image frame; the arm area, the head area, etc. Factors such as the shadow formed by the smoothness of the human hand should also be considered in the list

However, at present, there is no relevant research on the SVM model conversion of in-plane gesture rotation. Only the model training of the given gesture requires a lot of data collection work.

Images