【摘要】 针对水下和低光图像存在的对比度低和颜色失真等情况,提出一种面向水下和低光图像的复原及增强方法。预处理操作,实现图像的复原;采用改进后的直方图均衡化方法,提高色彩的饱和度和图像的清晰度;运用卷积神经网络结合非二次采样轮廓变换(NSCT)技术,解决曝光问题,避免图像色彩退化和边缘细节减弱的发生。实验结果表明,该方法增强了图像的色彩饱和度,保留了图像的边缘细节,具有较好的视觉效果。更多还原

【Abstract】 Aiming at the low contrast and color distortion of underwater and low-light images,a restoration and enhancement method was proposed for underwater and low-light images.The image was preprocessed to achieve image restoration.An improved histogram equalization method was adopted to improve color saturation and image clarity.The convolutional neural network was combined with non-subsampled contour transformation(NSCT)technology to solve the exposure problem and avoid the degradation of image color and the reduction of edge details.Experimental results show that the proposed method enhances the color saturation of the image and retains the edge details of the image,which has better visual effects.更多还原