基于散射模型的水下图像复原  

Scattering Model Based Underwater Image Restoration

作　　者：边信黔[1] 王晓娟[1] 

机构地区：[1]哈尔滨工程大学自动化学院,黑龙江哈尔滨150001

出　　处：《机器人》2010年第6期721-725,共5页Robot

摘　　要：水下图像的衰减度与成像距离有直接关系.为了在没有精确深度图情况下有效恢复水下图像,提出基于散射模型的分段映射方法.研究了水下光线的散射模型,探讨了水下图像的衰减规律;利用水池试验数据采用直线拟合法建立散射模型;提出基于散射模型的分段映射法.为避免求解精确深度图,利用映射约束条件及少量先验信息导出多级离散深度值,从而构建出从退化图像到真实图像的分段映射函数.恢复结果表明,所提方法增加了图像整体对比度,突出了图像细节,明显提高了图像质量.The attenuation of underwater image has a direct relation with the imaging distance.To restore the underwater image without accurate depth map,a segmented mapping approach based on scattering model is put forward.The scattering model of underwater light is studied and the attenuation law is explored on base of this model;the scattering model is built by linear fitting using pool test data;the segmented mapping approach based on scattering model is presented.To avoid finding out accurate depth map,discrete multiple-scale depth values are derived from mapping restrictions and a few prior knowledge.As a result,the function of segmented mapping from the degraded image to the actual image is built.Restoration results show that the approach is capable of enhancing the overall contrast,highlighting the detail,and improving the image quality effectively.

关 键 词：水下图像复原 散射模型 分段映射法 水下视觉 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]