Design And Research On The System Of Low-light-level Intelligent Surveillance Based On FPGA And DSP

As more attentions are paid to the security of life and work conditions, the system of Low-Light-Level intelligent surveillance is widely researched in military, medicine, security, industry, astronomy, etc.. The images are usually blurred because the dim light signal is easily corrupted by the noise, which becomes a bottleneck during the development of the Low-Light-Level intelligent surveillance system. To solve this problem, relative measures should be taken during different steps of the imaging process respectively.Based on the hardware design of the Low-Light-Level intelligent surveillance system, this paper mainly researches on the key technologies of imaging in order to get clear image, including auto exposure algorithm and auto focus algorithm. The main work is described as follows:1. Detailed analysis and design on the image acquisition driving system are made based on the selected image sensor ICX659ALA, a Low-Light-Level CCD. A precise method to adaptively adjust the exposure time of CCD is proposed and demonstrated depend on the design of the driving circuit and timing of ICX659ALA. CCD reading clock is taken as the unit to adjust the exposure time. The system's responsiveness and the environment adaptiveness were improved. FPGA is chosen as hardware design platform, and driving time is described with Verilog. The simulation in QuartusII and experimental results show that the designed driving time can precisely adjust the exposure time.2. Detailed analysis is made of the noise-resistant, accuracy and operation time in focus evaluation algorithms in the space domain. The solution to focus evaluation algorithm is concluded in the focused region and out focused region of the searching algorithm. The grey level difference function is improved. Comparison and analysis of the new focus evaluation algorithm and the original function are made. The simulation results show that the noise-resistant and accuracy are improved significantly, though the operation time increased slightly.