一种结合形态滤波和标记分水岭变换的遥感图像分割方法  

Segmentation of Remote Sensing Imagery Combined with Morphological Filtering and Marked Watershed Transformation

作　　者：江怡[1] 梅小明[1] 邓敏[1] 陈杰[1] 陈铁桥[1] 

机构地区：[1]中南大学地球科学与信息物理学院,湖南长沙410083

出　　处：《地理与地理信息科学》2013年第2期17-21,共5页Geography and Geo-Information Science

基　　金：国家863计划资助项目(2012AA121301);测绘遥感信息工程国家重点实验室开放基金(09R03)

摘　　要：分水岭变换是一种经典而有效的图像分割工具,能够得到单像素宽、定位精确且连续封闭的边缘轮廓。然而,由于图像背景噪声、地物内部细密纹理及参数设置不当等影响,直接对梯度图像进行分水岭变换容易产生过分割现象。为了克服这一问题,提出一种结合形态滤波和标记分水岭变换的遥感图像分割算法。首先,采用混合开闭重构滤波对原始图像进行去噪处理;然后,利用开闭重构对梯度图像进行形态学修正,并采用扩展极小变换运算对修正后的梯度图像进行极值标记;最后对其进行分水岭变换,得到分割结果。实验结果表明:该算法能有效抑制分水岭变换的过分割现象,通过调整滤波器尺寸和选取深度阈值参数,在一定程度上实现了快速而有效的分割,能够满足不同遥感图像分割要求。Watershed transformation is a classical and efficient tool for image segmentation,which can automatically generate a series of one-pixel wide,precisely located and continuous closed edges.However,it can cause over-segmentation in situations of gradient images with strong noise,quantity error and detailed texture.In order to overcome the inherent drawback,an algorithm combined morphological filtering with marked watershed transformation is proposed.Firstly,the original image is smoothed by morphological filtering;then,the gradient image is reconstructed by the opening closing filter and morphological marking;finally,the processed gradient image is segmented using the watershed transformation.Experimental results show that the proposed algorithm is successful in overcoming the problem of over-segmentation and can produce accurate segmentation result by adjusting filtering sizes and depth threshold,thus it is quite qualified for segmentation of different remote sensing imageries.

关 键 词：遥感图像分割 分水岭变换 形态滤波 形态标记 

分 类 号：P237.4[天文地球—摄影测量与遥感]