DNN group gesture identification method based on optimized gesture database distribution

Abstract

The invention provides a DNN group gesture identification method based on optimized gesture database distribution, wherein the method belongs to the field of a computer. The method comprises the steps of (1), acquiring gestures and forming a gesture data set; (2), re-classifying the gesture set by means of optimized gesture database distribution for obtaining a plurality of sub-databases; (3), performing learning training on a DNN model on each sub-database, and obtaining DNN structures; (4), inputting a to-be-identified gesture, acquiring the to-be-identified gesture by means of Kinect equipment, realizing gesture segmentation by means of a background subtracting method, and separating human hands from a background; (5), respectively transmitting the to-be-identified gesture to each DNN structure for identification, and calculating an output error E of each DNN identification result by means of a formula of E=(anticipated output)-(network response); and (6), returning an output result which corresponds with a least output error E, and determining the output result as the identified gesture.

Description

technical field [0001] The invention belongs to the field of computers, and in particular relates to a DNN group gesture recognition method based on optimized gesture library distribution. Background technique [0002] The bare-hand gesture recognition system includes static gesture (pointing gesture, single hand shape) recognition and dynamic gesture (pointing action, consisting of a series of gestures) recognition, which provides users with a more natural and direct way of human-computer interaction, and has gradually become a A research hotspot in the field of computer vision and human-computer interaction. Scholars put forward the Tortoise model to represent the basic features of the human hand, and then train the self-learning gesture pattern library based on it. Combined with the rule of survival of the fittest in the genetic algorithm, the target hand shape is realized in the geometric and texture hybrid feature space. It matches with the hand shape in the pattern lib...

AI Technical Summary

Problems solved by technology

[0002] The bare-hand gesture recognition system includes static gesture (pointer gesture, single hand shape) recognition and dynamic gesture (pointer action, composed of a series of gestures) recognition, which provides users with a more natural and direct way of human-computer interaction, and has gradually become a A research hotspot in the field of computer vision and human-computer interaction. Scholars proposed the Tortoise model to represent the basic characteristics of the human hand, and based on this, trained the self-learning gesture pattern library, combined with the rules of the survival of the fittest in the genetic algorithm, to achieve the target hand shape in the geometric and texture mixed feature space Matching with the hand shape in the pattern library and recognizing gestures improves the real-time performance, but only 10 kinds of gestures with a high degree of discrimination are selected for experiments; the researchers also extract the feature pixels of gesture edges, and use the idea of ​​Hausdorff distance template matching to realize Chinese Sign language letter recognition has a small amount of calculation and strong adaptability, but gesture recognition when gesture rotation, scaling, and skin color interference are not considered; some researchers use gesture shape features for gesture recognition, because the algorithm uses contour features. Gestures with very similar shapes; some studies use the density distribution feature (DDF) of images to realize the retrieval and recognition of binary images, but they cannot recognize gestures with less discrimination

Images