基于边缘与区域结合的光学遥感图像舰船分割方法

Method for Ship Segmentation in Optical Remote Sensing Images Based on Edge and Region Information

针对光学遥感图像中舰船目标分割精度低等问题,提出了一种基于边缘与区域结合的图像中舰船目标分割方法。首先,利用导向滤波改进Canny算法对待分割舰船遥感图像进行预处理,以保持边缘完整结构,得到舰船目标粗分割结果;然后,提出基于布谷鸟搜索优化的倒数灰度熵阈值选取方法进行舰船目标细分割;最后,引入形态学方法对分割结果进行修正,确保舰船目标的完整性。大量试验结果表明,该方法表现出较优的分割性能,为舰船检测识别领域提供了更好的分割方案。

Aiming at problem of low segmentation accuracy for ship target in optical remote sensing images,a ship target segmentation method in images based on edge and region information is proposed.Firstly,the traditional Canny algorithm is improved with guided filtering,and the ship's remote sensing image waiting for segmentation is preprocessed to maintain the edge integrity of ship's structure.Thus,a coarse segmentation result is obtained.Then,threshold selection method for reciprocal gray entropy based on optimized cuckoo search is presented to carry out fine ship target segmentation.Finally,mathematical morphology method is introduced,and the segmentation result is corrected.Thus,the complement of ship target is ensured.A large number of experimental results show that the method shows excellent segmentation performance,and it can provide a better segmentation scheme in domain of ship detection and recognition.