C++图像去雾算法研究+文献综述
关键词 图像融合 高斯金子塔 拉普拉斯金子塔 对比度增强 图像去雾
毕业论文设计说明书（论文）外文摘要
Title On Image Dehazing Algorithm
Abstract
The algorithm of dehazing images is studied in this thesis. First, the thesis reviews the state-of-the-art algorithms. Then, a detail algorithm of dehazing images is depicted. This algorithm is a multi-fusion based dehazing method which applys the Laplacian pyramid and Gaussian pyramid. The framework of the technique consists of three steps: First of all, the algorithm conducts the white-balance and contrast enhancement on the original hazy image separately and derives two different images. Then, in order to blend the information of the derived two images effectively and preserve the regions with good visibility, the algorithm filters their important features by computing three measures: luminance, chromaticity and saliency. At last, the algorithm uses a Laplacian pyramid representation in a multi-scale fashion to minimize artifacts introduced by the weight maps. The thesis tests forty images in C++ environment and the experimental results show that the algorithm performs well on thin haze but fails to deal with dense haze.
Keywords image fusion Gaissian pyramid Laplacian pyramid
contrast enhancement single image dehazing
目 录
1 绪论 1
1.1 课题研究背景及意义 1
1.2 现有的去雾算法介绍 2
1.3 本文的去雾算法及论文章节安排 3
2 基于多尺度图像融合的去雾算法 6
2.1 白平衡处理和对比度增强处理 6
2.2 依据测度权重筛选主要信息 11
2.3 多尺度融合 14
3 测试效果和结论 18
3.1 实验结果 18
3.2 结果分析 20
结 论 23
致 谢 24
参考文献25
1 绪论
1.1 课题研究背景及意义
在户外场景中,计算机视觉系统已经得到了广泛应用，如导航与制导、目标识别与跟踪、遥感、视频监控、无人驾驶汽车和无人机等。然而，目前的大多数系统鲁棒性不好，极易受到天气条件的影响。遇到雾、霾等天气时，悬浮在空气中的液滴、气溶胶以及固体微颗粒物会对光线产生折射，散射或吸收等作用，从而严重降低户外的水平能见度，导致成像系统采集的图像质量大幅度下降，这已经成为阻碍视觉系统广泛应用的重要因素[1]。
大气粒子对光线产生的作用主要有吸收和散射两种[2]。在近红外波段，主要为对光线的吸收作用；而在可见光范围内，大气粒子的散射作用占主导。随着成像系统到物体距离的增加，大气粒子对光线的散射作用，对成像系统的影响逐步增强。这种影响可分解为两个过程：⑴由于大气粒子的散射作用，部分大气光进入了成像系统。⑵在传输到成像系统的过程中，大气粒子对物体表面反射光的散射作用。作用的综合结果为采集到的图像偏色严重，对比度低，且拍摄到的远景模糊。
上一篇：基于图像的边缘检测算法研究