基于边缘保持中值滤波和BP神经网络的自然手势识别研究

Research on Natural Gesture Recognition Based on Edge Preserving Median Filter and BP Neural Network

旨在对噪声干扰下的自然手势进行提取并识别,通过图像小波去噪、提取边缘的中值滤波处理,再将手势样本二值化并进行区域特征的提取,得到自然手势的二值化图像。基于BP神经网络对自然手势模拟,实现自然手势的去噪识别。对已有自然手势识别技术进行分析并做出合理展望,为自然手势识别研究提供新思路。

Extracts and identifies the natural gestures under noise interference. Through the image wavelet denoising, the median fltering process of the edge is extracted, and the gesture samples are binarized and the regional features are extracted to obtain the binarized image of the natural gestures.Based on BP neural network for natural gesture simulation, to achieve the natural gesture denoising identifcation. Analyzes the existing natural gesture recognition technology and makes a reasonable prospect, and pruvides a new idea for natural gesture recognition research.