基于改进Prewitt算子的矿井视频图像小波阈值去噪

由于矿井成像环境的复杂性，导致所获取的图像较为模糊且存在一定程度的噪声，有必要采用适当的方法滤除其中的噪声并提高图像的清晰度。为此，提出了一种结合边缘检测的小波域矿井视频图像去噪算法。首先对噪声图像采用均值滤波算法进行预处理；然后对分别从检测模板、自适应阈值设定方法对经典的Prewitt算子进行适当改进，并采用改进的Prewitt算子提取预处理后图像的边缘轮廓信息，从而获得边缘图像和非边缘图像；最后对于非边缘图像，采用一种基于自适应阈值的改进小波阈值函数模型进行去噪处理，该模型充分结合了小波硬、软阈值去噪模型的优势，能够随着小波分解层数的变化自适应调整阈值大小。为了验证该算法的有效性，采用2幅实地获取的山东兖州某煤矿井下视频监控图像进行试验，并与小波硬、软阈值函数模型及2类已有的改进型小波阈值函数模型进行去噪效果对比。结果表明，该算法的去噪效果优于其余几类函数模型，且算法耗时略占优势，对于提高矿井视频监控图像的质量具有一定的参考价值。

Mine Video Image Wavelet Threshold Denoising Algorithm Based on Improved Prewitt Operator

Due to the complexity of imaging environment of mine，the acquired mine video image is relatively vague and blended with a amount of noise，so，it is necessary to take the appropriate method to filter out the noise and improve the resolution of the image.A new wavelet domain mine video image denoising algorithm based on edge detection is proposed so as to deal with the noise image.Firstly，the noise image is pre-processed by adopting average filtering algorithm；secondly，the classical Prewitt operator is improved from the perspectives of detection templates and adaptive threshold setting method，and the improved Prewitt operator is adopted to extract the edge profile information of pre-processed image to obtain the edge image and the non-edge image;finally，the non-edge image is filtered by the improved wavelet threshold function model based on adaptive threshold，the new function model is a combination of the advantage of wavelet hard and wavelet soft threshold function model and the threshold of the new function model can be changed adaptively according to the layer number of wavelet decomposition.In order to verify the effectiveness of the proposed algorithm，the two mine video surveillance monitoring system images obtained in a mine of Yanzhou，Shandong province are taken as the experiment data，the experimental results show that，the performance of the algorithm in this paper is better than the wavelet hard function model，wavelet soft function model and the two existed improved wavelet function model.Besides that，the time-consuming of the algorithm proposed is shorter than the two existing improved wavelet function model，and it also has some reference for enhance the quality of the mine video surveillance monitoring system images.