Optoelectronic Hybrid Joint Transform Correlator And Image Processing

With the rapid improvement of the technology and the level of industrialization, the technology of pattern recognition has been widely applied in these fields. It has been commonly used in the fields of the characters recognition, the industrial components recognition and the diagnosis, the remote sensing image recognition, as well as the military target recognition. As for now, the optic-electronic hybrid recognition technology has become the top of the list of the fast developing technologies due to the advantages of the programmable control and the high-speed parallel computation under huge amount of data. The extraction of the targets in the video image is the first step of the real-time joint transform correlation of the recognition system, meanwhile, the performance of the non-changing distortion is the most important index which tells the false targets apart from the true ones.This thesis discussed the basic theory of the optic joint converter based on DSP under the foundation of the optical matched filter correlation, and raised the technology which is modified and filtered towards the joint power spectrum based on the feature of poor performance of the classical joint transform correlation to identify.Then, according to the feature of the static gray-scale image, MATLAB has been used to deal the procession of the histogram equalization, such as image enhancement and image segmentation algorithm. The application scope of these image processing algorithms has been determined and programmed by C language, which used MFC as the platform. All these have formed great theoretic basics for the further experiment in order to subtract the target from the video images. The distortion question of the ratio rotation between the target image and the reference target has been conquered by the methods above.Finally, the CCS integrated development environment and the software development process of the DSP have been familiared. By programming the control and recognition codes of the real-time joint transform correlation identification system, every part of the system has been converged effectively. The functional modules have been transplanted into DSP such as the image pre-procession which makes the system more advanced by accomplishing the function of self-recognizing.