【作者】 李琼；

【导师】 许开宇；

【作者基本信息】 上海海事大学 ， 通信与信息系统， 2006， 硕士

【摘要】 基于红外图像的目标检测技术是现代光电子技术领域的重要组成部分。红外成像系统因为具有良好的抗干扰能力，能透过烟雾等障碍探测到目标，实现昼夜连续工作，所以红外检测系统可以作为雷达等探测系统的补充，在船舶避碰、海岸监视和交通管理中发挥重要作用。 本文在算法理论、系统软件设计和硬件实现等几个方面进行了研究。论文的第一部分对海面背景下的运动目标检测算法进行了研究，通过对运动目标检测算法的分析比较，提出了一种基于Sobel算子的帧间差分检测算法。传统的帧间差分法算法简单、便于硬件实现，但在信噪比较低时抗噪性能较差。而本文研究的中等距离船舶目标，与点目标相比具有一定的边缘特征，因此将边缘检测与帧间差分相接合便可获得良好的抗噪性能。文中对多种边缘检测算子进行了分析比较，其中Sobel算子在处理该类图像时的速度与效果比较理想，且算法简单易于DSP硬件实现，所以最终采用了sobel算子。本文第二部分分析了DSP在图像处理中的优势，详细介绍了其片内外设EDMA在系统中的应用，设计了基于TMS320C6711 DSP的红外船舶目标检测系统的硬件电路，并在该平台上实现了基于Sobel算子的图像检测算法。系统软件采用C语言编写，并根据C6711的硬件特征进行了优化，优化后运行效果良好。 由于应用的特殊性，本文在研究过程中还特别注意了算法的实时性，在软硬件配合优化后，算法实时性能达到应用要求。用DSP实现检测算法对于系统的小型化、便携化很有意义，因此，本文对于构建船舶目标实时检测系统有一定的价值。更多还原

【Abstract】 Target detection and recognition technology on infrared image is an important part in modern photoelectron technology field. With good jamming performance, infrared imaging system can detect target through such obstacles as smoke, dust, fog, etc. and realize continuous passive detection day and night. So, infrared detection and recognition system can serve as a supplement to radar and other detection and it will play an important role in ship collision prevention, ship navigation safety, and Vessel Traffic Service.This dissertation has finished algorithm analysis and has designed the ship target detection system. Algorithm has been realized on the software and hardware, and ship target can be detection accurately. The first part of dissertation studied the algorithms of moving target detection in sea-background. By analysis and comparison the algorithms, the dissertation presented an algorithm, based on the Sobel operator and frame difference method. Considering the characteristics of infrared image, the frame difference algorithm is simple and easy to be realized by hardware. But the algorithm’ anti-noise performance is not well while the signal-to-noise ratio is low. Compared with the point target, the moving ship target captured at intermediate range has certain edge characteristic. So by joining the edge detection algorithm and the frame difference the algorithm can obtain better anti-noise performance. By comparing several edge detection operators, it considered that the Sobel operator has higher processing speed and better performance and easier realized by DSP. In the second part, this dissertation analyzed the superiority of DSP in image processing and particularly introduced the application of EDMA in system. Based on the TMS320C6711 DSP,更多还原