基于多峰直方图的遥感图像阈值自动选取算法

An Automatic Threshold Selection Approach for Remote Sensing Imagery of Multimodal Histograms

基于多峰灰度直方图的阈值自动选取算法,首先使用平均滤波器平滑直方图,消除直方图中微小的波动,保留反映直方图总体趋势的大的波峰波谷;然后通过寻找直方图中局部最大值和最小值,以确定合理的阈值.该算法使用时仅需要用户指定直方图应保留的合理波峰数,在针对大尺寸遥感图象计算出的阈值要优于大津算法和双峰算法计算出的阈值,且算法计算复杂度较低,适用于快速阈值选取.通过对TM影像中水体自动提取实验研究,阈值自动选取算法可用于水体及其它地物目标遥感信息的批量自动提取处理,具有较好的提取效果.

An automatic threshold selection approach is enhanced for remote sensing imagery of multimodal histograms. Firstly, the mean filter is utilized to smooth the multimodal histogram, eliminate the small os- cillation in the histogram and keep the large scale peaks and valleys, which reflect the overall trend of the histogram. Secondly, the reasonable threshold is decided by searching the local maxima（peak） and local minima（valley） in the histogram. The algorithm only requires the user to determine the reasonable number of peaks, which should be kept in the histogram. For the large scale remote sensing imagery, the calculated thresholds are better than those calculated by Otsu method and bimodal method. Moreover, the computa- tional complexity of the algorithm is relative low and it is suitable for fast threshold selection. The experi- ment on automatic extraction of water body in TM imagery shows that the proposed algorithm can effec- tively and efficiently batch extract remote sensing information such as water body,alluvial fan,etc.