A Gesture Recognition Method Based on Random Projection Multi-kernel Learning

Abstract

The invention discloses a gesture recognition method based on random projection multi-core learning. The steps include: collecting gesture images, and preprocessing the images, the preprocessing includes gesture positioning and gesture segmentation; extracting SIFT features from the preprocessed and segmented gestures; Use the k-means algorithm to train the learning dictionary, and then use the iterative dictionary update algorithm to update the dictionary; divide the gesture image into a spatial pyramid, and encode the sift feature of the gesture image in each layer of the spatial pyramid according to the dictionary obtained from training, and obtain eigenvectors, and concatenate the eigenvectors, and then use random projection to reduce the dimensionality of the eigenvectors; learn the kernel matrix for the eigenvectors after dimensionality reduction in each layer of the pyramid, and use the multi-core model learning algorithm for classification learning to obtain the optimal kernel matrix combination factor. The invention solves the problems of background interference, high complexity, long time consumption, low recognition rate and the like existing in the current traditional gesture recognition method.

Description

technical field [0001] The invention relates to the technical field of pattern recognition, in particular to a gesture recognition method based on random projection multi-core learning. Background technique [0002] At present, with the continuous advancement of science and technology, human-computer interaction has developed rapidly, and human-computer interaction has become one of the hotspots of researchers. The goal of human-computer interaction is to realize natural communication between users and machines, and to provide users with real-time and intuitive interactive experience. Since information is transmitted between people through language, body and expression, and gestures are natural and intuitive, human-computer interaction based on gesture recognition has attracted more and more attention. play an increasingly important role. Gesture recognition involves multiple disciplines, such as computing science, machine learning, pattern recognition, image and video pro...

AI Technical Summary

Problems solved by technology

[0008] In order to overcome the shortcomings and deficiencies of the existing technology, the present invention provides a gesture recognition method based on random projection multi-core learning, which solves the problems of background interference, high complexity, long time consumption and low recognition rate in the current traditional gesture recognition method And other issues

Images