基于颜色校正及特征融合的海洋遗迹图像增强

Marine relics image enhancement based on color correction and features fusion

针对海洋遗迹图像出现色偏、细节模糊等现象,导致海洋测绘等工作效率低的问题,提出基于颜色校正及特征融合的海洋遗迹图像增强方法。首先改进传统颜色校正以缩小色彩衰减程度;其次使用深层信息提取模块获取细节信息,并加入多尺度特征提取模块保证图像视觉的完整性;最后提出特征融合模块融合深层细节和尺度特征信息,输出清晰海洋遗迹图像。实验结果表明:本文将传统方法和深度学习算法结合,不仅可以提高真实海洋遗迹图像的清晰度,而且在数据集上的评价指标均优于对比的经典及新颖算法,为海洋测绘等领域提供新思路。

In order to solve the problem of low efficiency of marine surveying and mapping caused by color deviation and blurred details in marine cultural relic images,a method of marine cultural relic image enhancement based on color correction and feature fusion is proposed.Firstly,the traditional color correction is improved to reduce the degree of color attenuation;Secondly,a deep information extraction module is used to obtain detailed information,and a multi-scale feature extraction module is added to ensure the visual integrity of the image.Finally,a feature fusion module is proposed to fuse deep details and scale feature information to output clear marine relic images.The experimental results show that the combination of traditional methods and deep learning algorithm can not only improve the clarity of real marine relic images,but also provide new ideas for marine mapping and other fields.