中文摘要：

在同一光源不同照射角度下，光变油墨数字颜色存在着明显的变化，这是进行纸币的真伪识别的重要方法之一，区别于人工识别，在纸币真伪鉴别上提出了针对光变油墨技术的机器识别方法，纸币机器自动识别系统主要由高速数字信号处理器（DSP）、现场可编程门阵列（FPGA）和彩色CCD摄像机组成。该系统由FPGA对视频数据流进行预处理并且控制多个存储区有序切换，完成纸币图像数据的实时采集存储；针对图像数据的特点，基于片上可编程系统（SOPC）的概念，FPGA中集成了Nios软核实现并行处理，DSP完成纸币的识别过程。通过实验验证了针对光变油墨技术的机器识别方法的可行性和正确性。

英文摘要：

The color of the figures printed with photo-induced discoloration ink has a special change when the light irradiates a bank note from different angles. This is an important method to identify if a bank note is faked or not. In this paper, an automatic recognition method focusing on photo-induced discoloration printed ink technique is proposed to recognize the fakeness of bank notes. The bank note automatic recognition system mainly consists of DSP, FPGA and color CCD camera. The system pre-handles video data stream by FPGA, sequentially switches among different memory areas and completes the real-time collection and storage of hank note image data. According to the characteristics of the image data, based on the concept of SOPC, FPGA integrates Nios soft core and implements parallel processing. Then DSP completes the recognition process of a bank note. The feasibility and effectiveness of the proposed automatic recognition method is verified by various experiments.