彩色图像的FPGA实时增强系统实现

Implement of system for color image enhancement in real-time based on FPGA

为了对彩色图像进行实时增强,本文提出了采用基于插值的分段拉伸算法。首先将彩色RGB图像转换为HSV空间,在该空间,对图像直方图进行基于差值的分段拉伸处理,以达到对图像增强的目的。经过该方法增强后的图像,细节信息更加丰富,图像的清晰度得到了改善,图像的视觉质量也得到了明显提高。经过该算法增强后图像的灰度平均梯度值为直方图均衡化算法的1.95倍。应用现场可编程门阵列(FPGA)为中央处理器,通过并行处理结构及流水线技术,完成图像空间的转换和图像的实时增强算法,简化了系统设计,使处理系统硬件更加紧凑,运行更加可靠。给出了系统主要功能模块的实现方法,经现场调试,可完成每秒30帧×1 024×1 024×24bit数据的处理,与直方图均衡化等传统图像增强算法相比,该算法计算时间缩短了0.807ms。该系统具有集成度高、图像处理速度快和实时性强等特点。

In order to enhance the color images in real time,this article proposed an algorithm of piecewise stretching based on interpolation.Firstly,the RGB color image is converted into HSV space.In this space,the histogram of the image is stretched based on interpolation.By using this method to enhance the image,the details of the image are more rich,the image clarity is improved obviously and the visual quality of the images has also been significantly improved.Gray average gradient value of enhanced image after this algorithm is 1.95 times than histogram equalization.In this paper,field programmable gate array （FPGA） is used as central processor.By using of parallel processing and pipelining technology,we completed the conversion of the image space and the algorithms of real-time image enhancement,simplified system design,and make the processing system more compact,the operation more reliable.The paper also gives the system functional modules.After debugging,this system could complete the processing of data 30 × 1 024 × 1 024 × 24 bit per second.Compared with the traditional image enhancement algorithms thus as the histogram equalization,this algorithm reduced the time about 0.807 ms.The system has the characteristics of high integration,fast image processing speed and strong real time.