摘要
基于单通道双谱夜视系统中对图像修复的需要,提出了一种基于水平集的图像修复及消噪技术。在阐述单通道双谱夜视系统工作原理的基础上,结合系统的实时性要求及待修复微光条纹图像的特点,设计了处理速度高的水平集修复算法;考虑微光图像的噪声对修复结果的影响,在修复的同时加入了热传导滤波方法。实验结果表明,该算法能够对微光条纹图像进行实时且有效的修复。
Based on the image inpainting need of single-channel dual-speetrum night vision system, an image inpainting and denoising technique based on the levelset was proposed. The levelset inpainting algorithm with high speed was designed by simply illuminating the working theory of single-channel dual-spectrum night vision system, its real time need and characteristics of waiting inpainting low light level stripe images. Considering the effect of noise on inpainting results, the method of heat conduction fihering was added to inpainting process. Experiments and results analysis have shown that low light level stripe images can be inpainted and denoised effectively by using the image inpainting and denoising algorithm based on levelset real-time.
出处
《计算机应用》
CSCD
北大核心
2007年第9期2114-2116,共3页
journal of Computer Applications
基金
国家自然科学基金资助项目(60572107)
关键词
微光条纹图像
图像修复
水平集
图像消噪
热传导方程
low light level stripe image
image inpainfing
levelset
image denoising
heat conduction equation
