大幅面光电扫描仪图像并行采集与校正拼接的设计和实现

发布时间：2018-08-11 17:30

【摘要】：随着现代工业生产中数字化程度的进一步加深，高精度海量数据采集系统在其中起到关键作用。传统低精度低速扫描由于对原始数据的采样率不够，容易造成采集图像变形，采集和处理速度缓慢，已经不能适应行业的发展需求。因此对原始数据采集的精细化，快速化作为数字处理前端，其发展趋势也显而易见，针对高分辨率和高速图像的采集也应运而生。大幅面扫描仪使用CCD线阵高分辨率高速扫描的技术手段已成为大势所趋。随着国外工业自动化技术的迅速发展，许多国外公司纷纷跟进并研发出满足特定需求的LFS(large format scanner)系统并占据了大量市场份额呈垄断状态，，由于国内对LFS系统的研究起步较晚，许多类型的LFS设备完全依赖于进口。因此，研究具有自主知识产权的大幅面光电扫描仪系统具有非常重要的意义。本文针对LFS系统中涉及的核心技术及其技术难点，详尽的从硬件和软件结构进行分析研究，从而提出了针对大幅面光电扫描仪适用的海量图像数据的并行采集方法和色彩校正算法，以更好的满足实际应用需求。主要内容包含：（1）大幅面光电扫描仪采集系统硬件模块设计，使用无频闪LED光学系统，配合DALSA高速采集卡与高分辨率CCD传感器采集并转换亮度信息得到原始图像信息；（2）并行采集系统采用多核处理器并行工作方式，用自建链式存储模型对内存进行管理，并动态同步管理多线程采集得到的图像数据，增强整个系统的稳定性，让采集效率得到提高；（3）应用颜色空间理论从图像产生的过程进行分析，提出了针对大幅面光电扫描仪适合的色彩校正算法，从而能够在短时间内完成颜色空间转化和颜色还原，使得整个采集系统的图像输出效果得到优化；（4）从机器视觉原理分析并完成SIFT特征点的提取算法，使得多幅图像的高精度拼接要求得到满足，大幅面扫描仪中使用SIFT算法进行拼接在国内未见报道。
[Abstract]:With the further deepening of digitization in modern industrial production, high-precision mass data acquisition system plays a key role in it. Traditional low-precision low-speed scanning can not meet the needs of the development of the industry because of the insufficient sampling rate of the original data, which is easy to cause image distortion and slow acquisition and processing speed. As the front-end of digital processing, the fine and fast acquisition of raw data has obvious development trend, and the acquisition of high-resolution and high-speed images has emerged as the times require.
With the rapid development of industrial automation technology abroad, many foreign companies have followed up and developed the LFS (large format scanner) system to meet the specific needs and occupied a large market share in a monopoly state, because of the domestic L-FS (large format scanner) system. The research of FS system started late, and many types of LFS equipment depend entirely on imports. Therefore, it is very important to study the large-area photoelectric scanner system with independent intellectual property rights.
In this paper, the core technology and its technical difficulties involved in LFS system are analyzed and studied in detail from the hardware and software structure. A parallel acquisition method and color correction algorithm for large-scale photoelectric scanner are proposed to better meet the practical application requirements.
(1) The hardware module of the acquisition system of large-scale photoelectric scanner is designed. The original image information is acquired by using non-stroboscopic LED optical system, DALSA high-speed acquisition card and high-resolution CCD sensor.
(2) The parallel acquisition system uses multi-core processors to work in parallel, uses self-built chain memory model to manage memory, and dynamically synchronizes the image data collected by multi-threading, which enhances the stability of the whole system and improves the acquisition efficiency.
(3) Color space theory is applied to analyze the process of image generation, and a suitable color correction algorithm for large-scale photoelectric scanner is proposed, so that color space conversion and color restoration can be completed in a short time, so that the image output effect of the whole acquisition system is optimized.
(4) By analyzing the principle of machine vision and completing the extraction algorithm of SIFT feature points, the requirement of high-precision mosaic of multiple images is satisfied. It has not been reported in China that large-scale scanner uses SIFT algorithm for mosaic.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP274.2;TP391.41

【参考文献】