基于多尺度分割的对象级影像平滑算法

An Object-oriented Image Smoothing Algorithm Based on the Convexity Model and Multi-scale Segmentation

提出了一种对象级的保边缘影像平滑算法。该算法利用空间聚类对影像进行多尺度分割,在分割过程中,提取出不同尺度下的符合凸面模型(convexity model)的影像对象(image object);依据对象的统计参数对影像对象进行筛选,符合要求的影像对象内部进行平滑处理,其余对象不受影响。利用该方法可以有效地去除噪声和无用小目标,在不破坏指定目标边缘的同时,实现影像的平滑处理。

In image process, smoothing process will effect object edges extraction. In remote sensing image application, it＇s one of the difficulties to implement smoothing image and ensure the accuracy of edge extraction for specific objects at the same time. In this paper, we propose an object-based and edge-preserve image smoothing algorithm which divides image into many image objects with multi scales by using spatial clustering. In the course of segmentation, all image objects which accord with the Convexity Model will be extracted. Based on the prior object-specified statistical parameters, only those meet the requirements will be smooihed and other ones will be unaffected. Without damaging the edges of specified objects, our method can effectively remove noises and insignificant objects.