无人机遥感山区公路图像去噪方法

为实现对遥感图像较好的滤波去噪效果并有效保持图像边缘细节,提出一种自适应开关均中值融合滤波方法,可有效处理无人机拍摄的公路遥感图像.首先,选定遥感图像的局部区域,并设定梯度变换阈值大小,同时为标记该区域内的各个像素点,定义标志数组.然后,与阈值进行比较,若当前用于比较的像素点的值比梯度阈值大,则进行自适应开关均中值融合滤波,利用自适应选定阈值来替代传统开关均中值滤波器的固定阈值;若当前像素点小于梯度阈值,则选取改进梯度倒数进行加权平滑处理,即针对图像局部区域建立相关函数,将权值参数按照局部统计特性适应性进行调整.由实验结果可知,与传统梯度倒数加权平滑算法相比,自适应开关内中值融合滤波算法能够滤除遥感图像中的脉冲和椒盐噪声,对其他类型的噪声也有较好的效果,同时保留了图像的边缘细节信息.实验结果表明,滤波处理的图像平均梯度指数提高了 ３．１６％ ,且参数均方误差( MSE )减少了约 ５％ ,具有较高的应用价值.

A Denoising Method of Remote Sensing Mountainous Road Image by UAV

In order to achieve a better filtering and denoising effect on vertical images and effectively maintain image edge details , an adaptive switch-average median fusion filtering method is proposed , which can effectively process road imaging taken by drones．The threshold size of the gradual transformation is set , and a flag array for marking each segment point in the area is defined．Then , it is compared with the threshold． If the value of the currently used comparison variable point is greater than the gradient threshold , then adaptive switch average median fusion filtering is performed , and the adaptive correction threshold is used to replace the fixed threshold of the traditional switch average median filter ; If the current conversion point is less than the gradient? threshold , the improved gradient reciprocal is selected for gradual reduction．The experimental results show that the image is related to the traditional gradient reciprocal conversion slope algorithm．The algorithm in this paper can filter out the pulse and salt and pepper noise in the quantized image．Other experimental results show that the average gradient index of the filtered image is increased by ３．１６％ , and the parameter MSE is reduced by a bout ５％ , which has higher application value．