Rapid gesture recognition method based on video frame characteristics

Abstract

The invention discloses a rapid gesture recognition method based on video frame characteristics, and the method specifically comprises the following steps of 1, preprocessing the video frame image data: preprocessing an input video frame, correcting the data, and mapping the data to different semantic spaces; 2, segmenting a standard gesture, carrying out threshold processing in a specific color space, reducing the noise by adopting the morphological transformation and a smoothing filter, and determining a gesture recognition area; 3, extracting the gesture characteristics, extracting the shape characteristics of the gesture area, including the Hu moment, Fourier descriptors and the like; 4, selecting the gesture recognition based on a learner and calculating the weight, selecting two different base learners, wherein the different base learners should be endowed with different weights; 5, weighting different gesture classification results by the integrated learning to obtain a final classification result. According to the method, the frame-level images are extracted from video data, the typical gesture features are extracted, and a higher precision recognition method is achieved through an integrated learning method.

Description

technical field [0001] The invention belongs to the technical field of video data intelligent analysis, in particular to a fast gesture recognition method based on video frame features. Background technique [0002] Gesture recognition is a way for computers to understand human body language, creating a richer bridge between machines and humans than text user interfaces or graphical user interfaces. Gesture recognition enables people to communicate with hardware devices and interact naturally without any mechanical devices. Gesture has become a hot direction of human-computer interaction due to its intuitiveness, naturalness and easy acquisition. [0003] Researchers in the field of gesture recognition mainly include: contact based on sensor data and non-contact based on vision. The sensor data-based method can obtain more gesture states and features more accurately, but wearable devices will affect the user experience; the vision-based method can enable the operator to in...

AI Technical Summary

Problems solved by technology

The sensor data-based method can obtain more gesture states and features more accurately, but wearable devices will affect the user experience; the vision-based method can enable the operator to interact with humans in a more flexible and natural way, but has its limitations , cannot guarantee the accuracy of gesture features with high quality and affect the recognition accuracy

Images