基于免疫抗体编码机理的卫星图像云探测

Investigation of cloud detecting for satellite images based on immune coding principle

针对卫星遥感图像中云与下垫面的复杂性和多样性,提出了一种新的在复杂背景下常用目标图像描述方法.通过类比生物免疫抗体特异性与其构成单元氨基酸性质的相关性,得到目标图像的免疫基元集合形式及其分类方法.借鉴生物免疫抗体编码顺序中氨基酸结合能量最小原则,统计分析出训练样本的免疫基元亲和度计算公式,实现了目标图像描述的有限多特征优化组合.对于云检测问题,最终提出云的免疫抗体编码方法,定义了云的免疫亲和度计算公式,并成功构建了云免疫抗体.经200幅IKONOS卫星图像测试,验证了该方法在识别率和运行时间效率方面的有效性.

To utilize the texture diversity of clouds and land in remotely sensed images, a novel describing method for image targets identification with complex background was proposed. The properties and classification of object image describing immune primitives was computed by the correlation between antibody property and the specificity of amino acid residues. The affinity formula of the training image＇s immune primitives was presented by statistical analysis, which bears an analogy with the lowest amino acids combinative energy according to the biological immune antibody coding principle, to achieve the finite dimension object image features＇ optimize combination. Furthermore, The methodology was employed in the cloud contamination area detection. The cloud antibody has been configured and the cloud antibody has been tested on 200 images and more than 97% of results were correct, which proofed the validity of immune describing method for object image recognition in complex background.