摘要
由于水下光照的特性和水体杂质,水下图像光照不均且存在大量的伪影与环境噪声.提出一种新的水下图像超分辨率算法,采用RGBD摄像机采集的深度图作为水下图像的引导,实现基于局部特征的超分辨率增强算法.首先,对深度图像应用改进的Canny算法获得场景边缘轮廓;其次对水下图像进行边缘轮廓对齐后修正部分边缘形成配对;最后,通过修改的联合双边滤波器对水下图像进行上采样,在超分辨率的同时提升边缘轮廓的对比度.结果表明,与现有的基于颜色模型的算法相比,通过轮廓引导的方法能增强水下图像的边缘特性,并且显著减少纹理拷贝所带来的伪像.
Due to the characteristics of underwater illumination and water impurities, underwater images are unevenly illuminated and there are a lot of artifacts and environmental noise. In this paper, a new underwater image super-resolution algorithm is proposed, which uses the depth map collected by the RGBD camera as the guidance of the underwater image to realize the super-resolution enhancement algorithm based on local features. Firstly, the improved Canny algorithm is applied to the depth image to obtain the scene edge contour;at the same time, the underwater image is aligned with the edge contour and corrected to form a match. Finally, the underwater image is upsampled by the modified joint bilateral filter to enhance the contrast of the edge contour in super-resolution. The experimental results show that, compared with the existing color model based algorithm, the edge characteristics of underwater images can be enhanced by the contour guidance method and the artifacts brought by texture copying can be reduced significantly.
作者
吴献
WU Xian(College of Computer and Cyber Security,Fujian Normal University,Fuzhou 350117,China)
出处
《福建师范大学学报(自然科学版)》
CAS
2022年第6期27-32,48,共7页
Journal of Fujian Normal University:Natural Science Edition
基金
福建省科技厅引导性项目(2019H0010)
福建省中青年教师教育科研项目(JAT170143)。
关键词
水下图像
图像增强
超分辨率
联合双边滤波
underwater image
image enhancement
super resolution
joint bilateral filter