基于自适应技术的鱼眼图像直接处理关键算法研究  

Study on the Key Algorithm of Fish-eye Image Directness Processing Based on the Auto-adapt Technology

作　　者：吴健辉[1] 张国云[1] 郭龙源[1] 邓己媛[1] 袁帅[1] 

机构地区：[1]湖南理工学院信息与通信工程学院,湖南岳阳414006

出　　处：《湖南理工学院学报（自然科学版）》2011年第3期35-40,共6页Journal of Hunan Institute of Science and Technology(Natural Sciences)

基　　金：湖南省教育厅科研项目(09C471);湖南省教育厅科研项目(10A046)

摘　　要：针对圆形鱼眼图像成像畸变大的特点,在现有数字图像处理算法的基础上,引入自适应技术开展了对鱼眼图像进行直接处理的相关关键算法研究.自适应技术通过对采集的图像画面进行分析,自动获取最佳的处理参数,从而针对不同的图像获得最佳的处理效果.论文首先从理论上对鱼眼图像直接处理的几个关键算法即模糊自适应中值滤波算法、自适应动态更新背景差运动目标检测算法和二维改进Otsu自适应鱼眼图像分割算法进行了详细的研究,然后对各个算法进行了实验仿真.实验结果表明,基于自适应技术的鱼眼图像直接处理算法能比较理想的对鱼眼图像进行去噪、运动目标提取以及目标分割,相对于传统的先校正后处理方法,具有运行速度快,信息无损的特点,在鱼眼超大范围智能视频监控中有着重要的实际意义.The circular fish-eye image is imaging by fish-eye lens and it has higher distortion because of its more than 180 degree field of view, so that it has bad processing effect if used some normal digital processing methods. In this paper, the auto-adapt technology was applied in the fish-eye image directness processing algorithm. Through analysis the image with automation, the best parameter would be obtained and the image would have the best processing effect to every image. The theory of key algorithm has been detailed researched and simulated for fish-eye image directness processing like Fuzzy Auto-adapted Median Filter （FAMF）, Auto-adapted Renewal Background Subtraction （ARBS）, and improved 2D Otsu algorithm for auto-adapted threshold image segmentation. The simulated result of these algorithms indicated that the directness processing algorithm for fish-eye image has higher processing speed and lossless information than traditional method of emend the fish-eye image to normal plane image. The processing effect is achieved the applied request practically and could be used in the super range video intelligent surveillance system.

关 键 词：自适应技术 鱼眼图像 直接处理 算法研究 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]