由于舰船行驶过程所拍摄图像容易受到各种外界因素影响，降低舰船图像质量，无法获取舰船重要信息，因此，有效利用多种视觉传达技术提升舰船图像质量。结合大气弹射模型，采用改进的四叉树搜索方法估计有雾舰船图像的大气光值，通过子窗口引导滤波器估计有雾舰船图像的透射率，反推大气散射模型，获取去雾后的舰船图像；通过视觉传达技术中低通滤波器将去雾后舰船图像分解为低频部分和高频部分，将舰船图像的低频部分输入到编码-解码深度网络中，修正舰船图像的颜色，将舰船图像的高频部分输入到残差网络中，增强舰船图像的清晰度；再经视觉传达技术中改进的双伽马函数调整舰船图像的亮度，实现舰船图像的质量提升。实验结果表明：该方法使舰船图像的纹理细节丰富，颜色鲜明，具有较高的清晰度和对比度，符合人眼的视觉感受；与原始舰船图像的结构相似性均在0.91以上，保证了舰船图像的真实性。
Due to the susceptibility of images captured during ship operation to various external factors, the quality of ship images is reduced and important information about the ship cannot be obtained. Therefore, various visual communication technologies are effectively utilized to improve the quality of ship images. Combining with the atmospheric ejection model, an improved quadtree search method is used to estimate the atmospheric light value of the foggy ship image. A sub window guided filter is used to estimate the transmittance of the foggy ship image, and the atmospheric scattering model is deduced to obtain the fogged ship image. By using low-pass filters in visual communication technology, the ship image after defogging is decomposed into low-frequency and high-frequency parts. The low-frequency part of the ship image is input into the encoding decoding deep network, and the color of the ship image is corrected. The high-frequency part of the ship image is input into the residual network to enhance the clarity of the ship image. By adjusting the brightness of the ship image through the improved dual gamma function in visual communication technology, the quality of the ship image is improved. The experimental results show that this method enriches the texture details of ship images, has bright colors, high clarity and contrast, and is in line with the visual perception of the human eye. After the application of this method, the structural similarity with the original ship image is above 0.91, ensuring the authenticity of the ship image.
2023,45(24): 200-203 收稿日期：2023-10-07
DOI：10.3404/j.issn.1672-7649.2023.24.038
分类号：TP75
作者简介：省明(1983-),女,博士,讲师,研究方向为视觉传达
参考文献：
[1] 马浩为, 张迪, 范亮, 等. 基于改进YOLOv5的雾霾环境下船舶红外图像检测算法[J]. 交通信息与安全, 2023, 41(1): 95-104.
MA Haowei, ZHANG Di, FAN Liang, et al. A Ship detection algorithm for infrared images under hazy environment based on an improved YOLOv5 algorithm[J]. Journal of Transport Information and Safety, 2023, 41(1): 95-104.
[2] 姚婷婷, 张波, 柳晓鸣. 特征增强全卷积网络下的船舶检测[J]. 计算机辅助设计与图形学学报, 2022, 34(7): 1028-1036.
YAO Tingting, ZHANG Bo, LIU Xiaoming. Feature enhanced fully convolutional network for ship detection[J]. Journal of Computer-Aided Design & Computer Graphics, 2022, 34(7): 1028-1036.
[3] 偰倩. 基于时空域滤波的视觉传达图像的微小细节增强方法[J]. 黑龙江工业学院学报(综合版), 2022, 22(10): 66-71.
XIE Qian. Micro-detail enhancement method for visual communication images based on spatial-temporal filtering[J]. Journal of Heilongjiang University of Technology (Comprehensive Edition), 2022, 22(10): 66-71.
[4] 翁鹏涛, 杜玉军, 张道奥, 等. 基于改进PGGAN的口腔图像数据增强算法[J]. 计算机工程与设计, 2022, 43(11): 3225-3234.
WENG Pengtao, DU Yujun, ZHANG Daoao, et al. Data augmentation algorithm of stemmatological images based on improved PGGAN[J]. Computer Engineering and Design, 2022, 43(11): 3225-3234.
[5] 上官小雨, 于跃波. 视觉传达的多帧平面图像纹理细节增强算法[J]. 吉林大学学报(信息科学版), 2023, 41(2): 359-366.
SHANGGUAN Xiaoyu, YU Yuebo. Algorithm of texture detail enhancement for multi-frame plane image based on visual communication[J]. Journal of Jilin University (Information Science Edition), 2023, 41(2): 359-366.
