车载手势识别中基于小波变换和双边滤波的图像去噪方法

发布时间：2018-02-23 21:14

本文关键词： 手势识别图像去噪小波变换多尺度双边滤波　出处：《北京理工大学学报》2017年04期 　论文类型：期刊论文

【摘要】：手势识别是人机交互中的重要研究领域,车载手势识别系统可以减少驾驶员手动操作仪表导致的分心,提高驾驶安全性.受光照变化、汽车环境、摄像头成像质量等各因素的影响,车载手势图像中常会存在大量复杂噪声.这些噪声严重影响后续手势分割、特征提取和手势识别的准确性.针对手势图像中存在的噪声问题,本文提出了一种适用于车载手势图像处理的新方法.该方法先对小波分解后的各高频子带采用不同方向的一维非线性扩散滤波处理得到初步去噪手势图,在此基础上用多尺度双边滤波对图像再次处理.实验结果表明,本文方法可以较好地去除车载手势图中噪声,抑制车载手势图细节的模糊.
[Abstract]:Gesture recognition is an important research field in human-computer interaction, vehicle gesture recognition system can lead to a reduction of manual operation of instrument driver distraction, improve driving safety. Affected by illumination changes, automobile environment, various factors affecting the imaging quality of the camera, due to the existence of a large number of complex vehicle gesture image noise. The noises seriously affect subsequent gesture segmentation, the accuracy of feature extraction and gesture recognition. Aiming at the problem of noise exists in gesture image, this paper proposes a new method suitable for vehicle gesture image processing. Firstly, the high frequency sub bands after wavelet decomposition of one-dimensional nonlinear diffusion filtering with different direction of preliminary denoising based on gesture map using multi scale bilateral filtering for image processing again. The experimental results show that this method can effectively remove the noise of vehicle gesture map, restraining the car The fuzziness of the details of the gestures.

【作者单位】：太原理工大学计算机科学与技术学院;
【基金】：国家自然科学基金资助项目(61540007,61373100) 虚拟现实技术与系统国家重点实验室资助项目(BUAA-VR-15KF02,BUAA-VR-16KF-13)
【分类号】：TP391.41;U463.6

【相似文献】