论文部分内容阅读
针对复杂地物背景下的城区红外遥感图像的云层干扰问题,提出了一种基于图像特征提取、区域投票表决和阈值分割的云层检测方法。对图像进行去噪和归一化拉伸处理,再进行多特征提取,并通过提取的特征向量对图像进行分区域云层投票表决,最后根据表决结果和分形特征度量矩阵进行阈值计算和阈值分割,并通过形态学处理得到精确的云层区域。统计检测结果显示算法对不同时刻的数据检测准确率在91%以上,证明了算法的适用性和有效性,为红外遥感图像的信息处理提供了有效的技术支持。
Aiming at the cloud interference of infrared remote sensing images in urban areas with complex objects, a cloud detection method based on image feature extraction, regional voting and threshold segmentation is proposed. The image is denoised and normalized, and then multi-feature extraction is performed. Then, the image is subjected to cloud voting by sub-region through the extracted eigenvectors. Finally, threshold calculation and threshold segmentation are performed according to the voting result and the fractal feature metric matrix. And through the morphological processing to get accurate cloud area. The statistical test results show that the accuracy of the algorithm for data detection at different time is above 91%, which proves the applicability and validity of the algorithm and provides effective technical support for the information processing of infrared remote sensing images.