Study On Image Processing System Of Joint Transform Power Pectrum Based On DSP

Before the DSP appeared, the nonlinear processing of image in opto-electronical hybrid pattern recognition mainly depends on the PC system or microprocessor (MPU). But processing speed of MPU is low; it can't satisfy the high speed and real-time requirement. On the other hand, the PC system is huge, and difficult to use.To meet the requirement of image processing of large quantity, high transmission ratio and complicated computation, the authors bring DSP into opto-electronical hybrid joint transform correlator, introduced a real-time image processing method of power spectrum based on the TDSDM642 platform. Experiments proved that the system can satisfy actual work requirement.IN this paper, the whole structure of opto-electronical hybrid joint transform correlator were discussed first, then the nonlinear processing of joint transform power spectrum (JTPS) were analyzed in detail. Third, the whole performance of power spectrum processing based on DSP and function of each part were analyzed. The paper gived the system structure frame diagram and software development processing, introduced the method of design and subject procedure of programme optimize. Based on the above theories, the power spectrum processing based on DSP was set up on DM642 EVM. Experiments proved that the proper pre-processing and binary of joint transform power spectrum can effectively enhanced the high frequency, resisted the noise, gained the sharper correlation peak and straitness correlation bandwidth and improved the ability of recognizing objects of the opto-electronical joint transform correlator. The image processing method of power spectrum based on DM642 is small, low power consume and high performance. Test on the spot and long-time running proved that the system is stable in operation, and satisfy the real-time processing requirement basically.