实时采集的光电视频图像受到成像环境、电子元器件特性及传感器材质等各种因素的影响，往往存在噪声污染严重和轮廓模糊的问题。为了降低图像的混合噪声、细化图像边缘，且满足实时需求，本文通过研究并改进图像预处理相关算法，设计一个基于FPGA的光电图像采集及预处理系统，具体研究内容如下：

针对传统滤波算法去噪单一和边缘检测算法精度较低的问题，提出一种改进的图像预处理算法：滤波方面在传统中值滤波及均值滤波的基础上，使用一种修正的阿尔法均值滤波算法，可以有效抑制混合噪声，改善传统滤波算法去噪的单一性；边缘检测方面使用一种形态学边缘检测算法，在传统Sobel算子的基础上，结合形态学腐蚀膨胀算法，更好地提取并细化图像边缘。改进算法通过MATLAB平台加以验证，实验结果表明，改进算法较传统算法去噪及边缘检测的效果更好。

搭建实时视频图像采集系统。系统包括图像采集模块、数据存储模块以及图像显示模块。图像采集模块选用CMOS摄像头采集实时图像；数据存储模块选用SDRAM芯片进行数据缓存，并调用SDRAM控制器IP核进行相关控制；图像显示模块调用视频输出IP核实现VGA时序，完成预处理后图像的实时显示。

对改进后的图像预处理算法进行优化设计，移植于FPGA平台实现。其中，修正的阿尔法均值滤波主要包含边界点填充、窗口生成、全比较排序以及加法树求和等子模块；形态学边缘检测主要包含梯度计算、阈值比较、腐蚀、膨胀等子模块。将设计好的图像预处理算法依次加入实时视频图像采集系统，实现光电图像的采集及预处理。

系统完成后上板进行结果验证。对各模块进行编译综合，逐层叠加算法，观察实验结果。实验结果表明，本系统有效抑制了噪声且细化了图像边缘，处理后图像的峰值信噪比提升了2.42dB，且每帧图像的处理时间达到约0.033s。经过系统分析得出，本文系统处理效果较好且具有实时性，在实时图像处理中具有一定的参考价值。

Due to the influence of several factors such as the imaging environment, the properties of electronic components and sensor materials, real-time photoelectric video images are usually affected by severe noise pollution and blurry outlines. In order to reduce image mixed noise, sharpen image edges, and fulfill real-time requirements, an FPGA-based photoelectric image capture and preprocessing system is constructed by researching and developing preprocessing algorithms. The concrete research contents are as follows:

 Aiming at the problems of single denoising of traditional filtering algorithms and low precision of edge detection algorithms, an improved image preprocessing algorithm is proposed：In terms of filtering, a modified alpha mean filtering method is used, which can effectively suppress mixed noise and improve the singleness of classic filtering algorithms for denoising. It is based on traditional median filtering and mean filtering；In terms of edge detection, a morphological edge identification approach is used, which can better extract and refine the image edge using the classic Sobel operator in combination with the morphological erosion and expansion technique. The improved algorithm is verified by the MATLAB platform, and the experimental results demonstrate that it outperforms the traditional technique in denoising and edge identification. 

 Build a system for capturing video images in real time. An image acquisition module, a data storage module, and an image display module are all part of the system. The image acquisition module chooses a CMOS camera to capture real-time images; the data storage module chooses SDRAM chips for data caching and calls the SDRAM controller IP core for related control; and the image display module calls the video output IP core to achieve VGA timing and completes the real-time display of the image after preprocessing.

 Improve the image preprocessing algorithm's design and port it to the FPGA platform. The modified alpha mean filter, for example, consists mostly of sub-modules like boundary point filling, window generation, complete comparison sorting, and addition tree summation, whereas morphological edge detection consists primarily of sub-modules like gradient computation, erosion, and expansion. To realize photoelectric image acquisition and preprocessing, the developed image preprocessing method is integrated to the real-time video image acquisition system.

After the system is completed, verify the system acquisition results. Compile and synthesize each module, layer by layer superimpose the algorithm, and test the results. The experimental results show that the system proposed in this study efficiently suppresses noise and refines the image’s edge, increases the peak signal-to-noise ratio of the processed image by 2.42dB, and processes each image frame in roughly 0.033s. The processing impact of the system in this work is good and real-time, and it has a particular reference value in real-time image processing, according to a systematic examination.

[1]杨勇奇,王典洪,徐朝玉,等.基于FPGA和USB的面阵CCD图像采集系统设计[J].现代电子技术, 2018, 41(16): 22-25+30.

[2]Park S Y, Lee K, Song H, et al. Simultaneous imaging and energy harvesting in CMOS image sensor pixels[J]. IEEE Electron Device Letters, 2018, 39(4): 532-535.

[3]甄好,王连明.一种基于多摄像头的大场景远程实时监控系统[J].东北师大学报(自然科学版), 2021, 53(04): 62-67.

[4]高璇,毕晓君.基于改进Faster R-CNN的低剂量CT图像肺结节检测[J].计算机应用研究, 2020, 37(S2): 404-406+409.

[5]邓鑫,陈紫强,郭朦,等.基于光照补偿的夜间车道线检测方法[J].激光杂志,2022,43(02): 71-76.

[6]邬亚胜.基于PYNQ软硬结合的二维手势交互设计[D].大连:大连理工大学,2021.

[7]吴厚航.深入浅出玩转FPGA[M].北京:北京航空航天大学出版社, 2010: 3.

[8]赵陆,文建平,莫为,等.图像处理算法IP核的异构验证框架[J].液晶与显示,2021, 36(07): 1042-1050.

[9]Chen P, Yu S, Chen B, et al. Design and SOPC-based realization of a video chaotic secure communication scheme[J]. International Journal of Bifurcation and Chaos, 2018, 28(13): 1850160.

[10]杨康,卫敏,孙磊.基于FPGA的自适应阈值边缘检测系统设计[J].传感器与微系统, 2019, 38(12): 96-99.

[11]夏润秋,刘洋.基于FPGA的PCI总线红外图像采集系统设计[J].液晶与显示, 2018, 33(09): 772-777.

[12]Khedkar A A, Khade R H. High speed FPGA-based data acquisition system[J]. Microprocessors and Microsystems, 2017, 49: 87-94.

[13]熊光阳,王野,李志茹,等.基于FPGA的千兆UDP/IP协议栈的实现及其在高速图像传输中的应用[J].仪器仪表用户, 2020, 27(03): 38-41.

[14]潘青松,张怡,杨宗明,等.基于Zynq的图像角点及边缘检测系统的设计与实现[J].计算机科学, 2017, 44(S2): 530-533+556.

[15]王玉晶,刘鑫.不同光照条件下基于二元模式的图像人脸识别[J].激光杂志,2020, 41(12): 151-155.

[16]Zhao H, Li H, Maurer-Stroh S, et al. Supervised segmentation of un-annotated retinal fundus images by synthesis[J]. IEEE transactions on medical imaging, 2018, 38(1): 46-56.

[17]Wang G, Liu Y, Xiong W, et al. An improved non-local means filter for color image denoising[J]. Optik, 2018, 173: 157-173.

[18]Vijaykumar V R, Mari G S, Ebenezer D. Fast switching based median–mean filter for high density salt and pepper noise removal[J]. AEU-International journal of electronics and communications, 2014, 68(12): 1145-1155.

[19]Xie X, Ge SL, Xie MY, et al. An improved industrial sub-pixel edge detection algorithm based on coarse and precise location[J]. JOURNAL OF AMBIENT INTELLIGENCE AND HUMANIZED COMPUTING, 2020, 11(5): 2061-2070.

[20]许四祥,李天甲,翟健健,等.基于Prewitt算子的自适应弱小目标检测[J].红外技术, 2019, 41(02): 189-193.

[21]Ravivarma G, Gavaskar K, Malathi D, et al. Implementation of Sobel operator based image edge detection on FPGA[J]. Materials Today: Proceedings, 2021, 45: 2401-2407.

[22]代文征,杨勇.基于改进高斯-拉普拉斯算子的噪声图像边缘检测方法[J].计算机应用研究, 2019, 36(08): 2544-2547+2555.

[23]张振海,尹晓珍,任倩.基于自适应均值的地铁隧道裂缝图像滤波算法[J].重庆交通大学学报(自然科学版), 2019, 38(06): 1-5.

[24]Dabhade S D, Rathna G N, Chaudhury K N. A reconfigurable and scalable FPGA architecture for bilateral filtering[J]. IEEE Transactions on Industrial Electronics, 2017, 65(2): 1459-1469.

[25]Alibabaie N, Latif A M. Self-learning based image decomposition for blind periodic noise estimation: a dual-domain optimization approach[J]. Multidimensional Systems and Signal Processing, 2021, 32(2): 465-490.

[26]罗朝阳,张鹏超,姚晋晋,等.一种基于形态学的边缘检测算法[J].计算机应用与软件, 2020, 37(02): 177-181+247.

[27]陈顺,李登峰.融合Canny算子和形态学的齿轮图像边缘检测研究[J].机电工程, 2020, 37(07): 821-825.

[28]尚卓,田玉.光电传感器噪声自适应控制系统设计[J].激光杂志,2020,41(11): 149-152.

[29]Fu B, Zhao X, Song C, et al. A salt and pepper noise image denoising method based on the generative classification[J].Multimedia Tools and Applications, 2019,78(9): 12043-12053.

[30]曾建华,黄时杰.典型图像边缘检测算子的比较与分析[J].河北师范大学学报(自然科学版), 2020, 44(04): 295-301.

[31]刘帆,强发军,赵明旭,等.图像质量客观评价方法研究与实现[J].数字技术与应用, 2020, 38(06): 107-110.

[32]Dev R, Verma N K. Robust noisiness measure based improved generalized fuzzy peer group for removal of mixed noise from color image[J]. IEEE Signal Processing Letters, 2018, 26(2): 267-271.

[33]祁欣,陈剑锋,罗伟林.图像预处理算法的FPGA实现[J].国外电子测量技术,2021, 40(02): 102-107.

[34]Heriana O, Praludi T, Wael C B A. Binary Template Matching for Morphological Dilation Enhancement in Navigation Radar Imaging[J]. Jurnal Elektronika dan Telekomunikasi, 2018, 18(2): 60-66.

[35]徐江涛,程思璐.智能化成像与识别CMOS图像传感器技术[J].航天返回与遥感, 2021, 42(06): 95-110.

[36]李明峰,周西峰,郭前岗.基于SOPC的视频监测系统[J]. 南京邮电大学学报(自然科学版), 2020, 40(01): 97-101.

[37]韩众,李智中,曹飞.基于FPGA的图像接口的设计[J].现代工业经济和信息化, 2018, 8(05): 37-39+97.

[38]Ecco L, Ernst R. Tackling the bus turnaround overhead in real-time SDRAM controllers[J]. IEEE Transactions on Computers, 2017, 66(11): 1961-1974.

[39]Arun G, Chaitra N, Divin Ganapathi T. Design and Implementation of VGA, Mixer and Phase Locked Loop[J]. International Journal of Innovative Technology and Exploring Engineering, 2020, 9(9): 2278-3075.

[40]Zhou J, Yao J, Zhang W, et al. Multi-scale retinex-based adaptive gray-scale transformation method for underwater image enhancement[J]. Multimedia Tools and Applications, 2021: 1-21.

[41]于天河,赵树梅,兰朝凤.结合视觉特性的红外图像增强方法[J].激光与红外, 2020, 50(01): 124-128.

[42]任勇峰,武昊男,储成群,等.基于FPGA的二维FFT图像边缘增强设计[J].图学学报, 2019, 40(01): 137-142.

[43]Lam H M, Guo F, Qiu H, et al. Pseudo Multi-Port SRAM Circuit for Image Processing in Display Drivers[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2020, 31(5): 2056-2062.

[44]Kumar S, Singla P. Sorting using a combination of Bubble Sort, Selection Sort & Counting Sort[J]. International Journal of Mathematical Sciences and Computing, 2019, 5(2): 30-43.

[45]师廷伟,金长江.基于FPGA的并行全比较排序算法[J].数字技术与应用, 2013, (10): 126-127.

[46]唐普英,姜书艳,张鹰.面向FPGA的16位加法器优化设计研究[J].工业和信息化教育, 2020(08): 91-94.

[47]杜建宝,董恩增,张祖锋.基于形态学的边缘检测系统与FPGA实现[J].天津理工大学学报, 2018, 34(04): 26-29+39.