摘要
Retinex图像增强算法能提升原图像的亮度、对比度和清晰度,但其运算比较复杂,难以满足许多工程应用所需要的实时性。通过采用SingLeton结构的高效浮点FFT,合并算法的多个循环体,合理安排指令并行和软件流水等措施大幅提高运算速度,并对图像进行分块处理以解决片内存储空间不够的矛盾,在单片TS201 DSP平台上实时实现了Retinex图像增强算法。实验结果表明该算法在时钟频率500 MHz的条件下,对512×256像素图像的处理用时18.5 ms。
The Retinex is an image enhancement algorithm that improves the brightness,contrast and sharpness.Many potential applications require the use of Retinex processing at video frame rates.It is difficult to achieve because the algorithm contains a large number of complex computations and data transfers.Accordingly,a fast computation algorithm executing on TS201 is proposed.The algorithm adopted series optimizations,such as SingLeton structure,paralleling of assembly instructions,loop unrolling and software pipelining.The input image is segmented in four blocks because of the limited memory of TS201.The experiment results prove that the algorithm can process a 512×256 image in 18.5 ms operating at a clock rate of 500 MHz.
出处
《激光与红外》
CAS
CSCD
北大核心
2012年第2期200-204,共5页
Laser & Infrared
