基于BP神经网络雾天图像复原算法.pdfVIP

您的论文得到两院院士关注 软件时空 文章编号:1008-0570(2011)02-0165-03 基于 神经网络的雾天图像复原算法 BP A Restoration Method for Fog Image Based on BP neural network ( 四川大学) 高艺文 贺可鑫 陶青川 李 柯 GAO Yi-wen HE Ke-xin TAO Qing-chuan LI Ke 摘要 针对雾天图像的特点 提出了一种基于粒子群优化算法的 人工神经网络的雾天图像复原算法 算法无需依据大气 : , BP 。 模型 它利用 神经网络的自学习 自记忆和泛化能力先用一组样本图像对网络进行训练 建立雾天图像与其对应的清晰 , BP 、 , , 图像之间的非线性映射关系 然后利用训练好的 神经网络对待复原的雾天图像进行复原处理 实验表明该算法能有效 , BP 。 地提高图像清晰度和对比度 视觉效果明显改善。 , 关键词 雾天图像 图像复原 神经网络 粒子群优化算法 : ; ; ; 中图分类号: TP391.41 文献标识码: A Abstract: In this paper, a new method is proposed for image restoration for fog image based on the Partical Swarm Optimization BP neural network. It does not need atmospheric mode, and the nonlinear mapping relationship between the fog image and its clear im- age is established by training the BP neural network which has the ability of learning, remembrance and generalizing with a group of 技 sample images. Then fog image which needs restoring could be restored by the trained neural network. The experimental results show that the restored images are clearer and have high contrast and the visual quality has been improved significantly. 术 Key words: Flog Image; Image restoration; Neural network; Partical Swarm Optimization; 创