基于USB3.0与FPGA的图像采集处理系统研究

发布时间：2018-03-28 23:12

本文选题：集成电路技术　切入点：图像采集　出处：《深圳大学》2017年硕士论文

【摘要】：图像信息具有直观形象、处理精度高等特点,其应用越来越广泛。与此同时,人们对于图像采集的质量和效率也有着更高的要求,这就需要图像采集处理系统解决一些现实存在的问题以满足这些需求。如今的图像采集逐渐向着高分辨率、高帧率发展,由此带来了图像数据传输效率和稳定性的问题,且随着图像处理复杂度的日益提升,图像处理在实时性方面也有所欠缺。本研究主要就是针对这两方面进行的。高分辨率、高帧率意味着需要高带宽。为了获取高速的传输速率,本研究选择的是基于USB3.0与FPGA的硬件架构。通过对数据传输方案进行深入分析,选择合适的数据传输类型和传输接口,实现USB3.0数据吞吐量的最大化,以提升数据传输的效率;利用帧校验的方式实现图像数据的帧同步,使得系统具备容错校正机制,并通过添加数据缓存模块以避免数据丢失,从而提升数据传输的稳定性。而在处理实时性方面,利用FPGA器件并行处理的特点,在图像采集的过程中添加图像预处理功能,此处主要讨论了图像卷积运算和统计排序运算的FPGA实现,并给出Sobel算子和中值滤波这两个相关的运用实例,说明硬件加速的实现,这可在实际的图像应用中,减轻后端图像软件处理的压力,从而减少图像处理的总时间。通过上述的方案,实验结果表明,该系统可稳定地进行2592×1944@14fps的图像采集,具备图像帧自动纠错功能,数据传输带宽可达320 MB/s,图像预处理功能工作正常。
[Abstract]:Image information has the characteristics of visual image, high processing precision and so on, and its application is more and more extensive. At the same time, people have higher requirements for the quality and efficiency of image collection. This requires the image acquisition and processing system to solve some practical problems in order to meet these needs. Nowadays, image acquisition is gradually developing towards high resolution and high frame rate, which brings about the problem of the efficiency and stability of image data transmission. With the increasing complexity of image processing, image processing is also lacking in real-time. High frame rate means high bandwidth. In order to obtain the high transmission rate, the hardware architecture based on USB3.0 and FPGA is chosen in this study. Through the in-depth analysis of the data transmission scheme, the appropriate data transmission type and transmission interface are selected. In order to maximize the throughput of USB3.0 data and improve the efficiency of data transmission, the frame synchronization of image data is realized by frame check, which makes the system have fault-tolerant correction mechanism, and by adding data cache module to avoid data loss. So as to improve the stability of data transmission. In the aspect of real-time processing, using the characteristics of parallel processing of FPGA devices, the image preprocessing function is added in the process of image acquisition. This paper mainly discusses the FPGA implementation of image convolution operation and statistical sorting operation, and gives two related examples of Sobel operator and median filter to illustrate the implementation of hardware acceleration, which can be used in actual image application. By reducing the pressure of back-end image software processing and reducing the total time of image processing, the experimental results show that the system can acquire 2592 脳 1944@14fps images stably, and has the function of automatic error correction of image frames. Data transmission bandwidth can reach 320 MB / s, image preprocessing function works well.
【学位授予单位】：深圳大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】