非局部均值图像降噪的嵌入式系统实现

摘要

Non - local means filtering is an effective image denoising technique. However, conventional algorithms generally require large storage space and high clock rates,therefore unsuitable for the implementation on an embedded system. In this study,a simplified non - local means denoising algorithm is described,which significantly reduces computational complexity with little loss of image quality. Image acquisition,non -local means denoising,and image display are implemented on an embedded system using S3C6410 processor based on ARM 11 and Windows CE. Both real - time operation and miniaturization are realized. Experiments show the system has satisfactory denoising performance.%非局部均值滤波是一种有效的图像降噪技术,但常规实现方法计算复杂,对存储空间和CPU频率要求较高,不适合用于嵌入式系统.面向嵌入式系统简化非局部均值降噪算法,以很小的图像质量损失为代价大大降低了计算复杂度.在基于ARM11处理器和Windows CE的系统上实现了图像采集、简化非局部均值降噪、显示等功能,可以满足图像处理系统实时性和小型化需求,降噪效果良好.