Font Size: a A A

Research On Scanning-photoelectric Warning System Based On Real-time Image Processing

Posted on:2012-12-07Degree:DoctorType:Dissertation
Country:ChinaCandidate:Z J GuoFull Text:PDF
GTID:1118330368995729Subject:Mechanical and electrical engineering
Abstract/Summary:PDF Full Text Request
In modern defense system and military struggle, the powerful security alert system is the foundation of a country. Currently, radar as the main facility performs various vigilance tasks. It has many advantages such as long detection distance, all days, every time of day and so on. But with the development of the current level of technology and the anti-radar equipment, radar has serious defects in the face of electronic interference, anti-radiation missiles, invisible aircraft and low-altitude penetration. The main reasons which caused these defects has several aspects:First, since the needs of the working principle of radar, it is required to transmit high power electromagnetics wave to probe the enemy. But the wave is also easy to be probed and disturbed by the enemy, and it even lead radar to be hit; Second, radar needs to receive the reflected waves to probe target. But with the development of various high-tech materials which can absorb electromagnetic wave by the surface of the target, the effect of radar is limited while facing this kind of invisible target; Moreover, because of the limit of ground clutter and the size of beam coverage area, radar is blind sometime.As a active detection equipment, it is the most difficult to overcome the shortcomings of radar. Countries in the world started looking for other ways to compensate for the shortcomings of radar. The passive optical alert system which has a passive detection equipment has a wide range of concern and attention. It uses a method based on optical image processing to achieve the early warning task. Its core is based on pattern recognition of digital image. In the passed 30 years, the technology of digital image data processing had rapid development, and it has been widely applied in various fields. The emergence of a large number of excellent processing methods and fast speed, lightweight embedded microprocessors, some other large-scale integrated circuits provides a very good foundation for the development of passive optical early warning system.In this paper, the scanning-photoelectric warning system based on real-time image processing is the main study object and the purposes of this papers are:First, a target recognition algorithm based on digital image are discussed and studied which provides theoretical basis for the system to realize its recognition task; The second is that a specific embeded hardware platform is researched and designed in order to complete the real-time image processing task of scanning-photoelectric warning system for target recognition.To achieve a large pitch angle and to get the precise point of target and to facilitate processing of image data in real-time, it uses a linear array CCD camera to collecte image data in real time; It proposes a classification algorithm based on rough sets and fuzzy neural network to achieve target recognition in digital image; By the same time, for obtaining a more simple neural network to improve the identification performance in the real-time, a ant colony algorithm is proposed to select the number of discretization intervals of continuous attributes; Large pitch angle and high-speed CCD camera brings a large number of image data. In order to process these image data in real-time a embedded hardware processing architecture based on 3 DSP+ FPGA is presented, and the PCB circuit board is drew; Under windows XP a transmission design in the system Based on DM642 PCI by the primary mode is completed. The design provides original image data and recognition results with a guarantee for the transmission in real-time.In the process of derivation of algorithms, sample training and the design of embedded real-time image processing hardware, we carry out a number of experimental tests to verify the correctness and feasibility. Finally, the experimentals based on aircraft targets which were collected by this system are assessed and analyzed. Experimental results show that the algorithm and hardware design of this system meet the early warning needs and it achieves the desired purpose.
Keywords/Search Tags:Passive warning, Linear CCD, Rough set, Neural network, Ant algorithm, Digital signal process(DSP), Target recognition
PDF Full Text Request
Related items