【摘要】 针对海上船舶目标的识别问题，提出基于光学遥感影像的船舶目标识别方法。利用高分卫星遥感资料为数据源，选取高分遥感影像为对象进行图像增强、裁剪和图像切片等操作。在切片图像中进行局部阈值分割处理，研究最大类间方差法(Otsu)、最大熵值算法(Kapur, Sahoo and Wong, KSW)和基于遗传算法(Genetic Algorithm, GA)的KSW算法等3种分割算法的性能差异。将局部阈值分割后的图像合成后，通过轮廓特征提取完海上船舶目标的识别。结果表明：在遥感影像的分割阶段，采用基于GA的KSW算法具有分割速度快，效率高且性能良好的特点，提出的船舶目标识别方法准确度高，对海上船舶目标的识别具有重要意义。更多还原

【Abstract】 The image processing technologies for recognizing ship targets from the data produced by high resolution remote sensing satellites are presented. Image enhancement, cutting and slicing are performed first. The sliced image is segmented by means of local threshold segmentation. Several segmentation algorithms are investigated, including Otsu method, the maximum entropy method and GA-based maximum entropy method. Ship targets are finally recognized from segment synthesized image through extracting the contour feature. The advantage of GA-based maximum entropy method in speed and effectiveness is demonstrated by experiments.更多还原