The Research And Impliment Of Infrare VideoDetection System Based On DAVINIC

Because of the limited distance between the streetlight and the car light beam, also because when the driver drives at night, he can easily get tired, so the traffic accident such as the pedestrian may be bumped too easily. As part of the vehicle auxiliary driving, the pedestrian detection of vehicle infrared images can remind the driver of the pedestrian position from very faraway, therefore he can judge in advance, so the probability of the traffic accidents may be reduced greatly to some extent.The pedestrian detection of vehicle infrared images system is using the passive infrared camera to capture the outside image. The hardware platform is based on the DM6437chip of the DAVINIC series of TI Company, and the software uses the BIOS operating system and it can achieve the transplant of the pedestrian detection of infrared images algorithm on the DM6437. Based on the DSP’s internal system flow, the algorithm is optimized maximally and to ensure the system’s real time operation which basically meets the actual usage requirements.The following research and design are introduced in this thesis:(1)Demand analysis and hardware strueture,Analyze several plans of the pedestrian detection of vehicle infraredimages system and compare their respective advantages and disadvantages. One innovation point of this system is to apply the plan of hardware DSP plus the software algorithm and then analyze the mainstream DSP’s advantages and disadvantages. Aiming at the requirement of this system, the DM6437is used as the main chip, analyze the in-out system of DM6437video and according to the system requirement, design the video coded module interface including the TVP5146introduction and the reason of choosing it as well as its function; the interface analysis of the outside storage module DDR2, the necessity and feasibility of using DDR2chip. According to the transmission speed and size etc, choose the appropriate memory with DDR2interface and achieves the interface design of this chip and DM6437;FLASH interface design; the necessity analysis of the extended FLASH storage, the EMIF controller and its interface design with specific FLASH chips; I2C/CAN interface analysis.(2)The design of hardware and software system. Software system analysis and design, mainly about the realization of the software development of the overall system, including the CCS3.3development environment building; adding VLIB bank to CCS3.3; the introduction of the basic platform BIOS operating system realized by the software and configure BIOS system to this system; realize the software structure building by API function provided by BIOS,Briefly introduce the realization of the pedestrian detection of infrared images including the algorithm of infrared pedestrian image segment, identification and pedestrian tracking algorithm,transplant pedestrian detection of infrared images algorithm based on the software structure; and finally realize the software coding debugging.(3)Mainly realized the optimization of the pedestrian detection of infrared images algorithm to DM6437including the principle of the software pipelining; the compiler optimization; the optimization through upgrading code locality; optimization based on changing data type;optimization based on numerical operation; optimization of reducing function call and jump instruction; adopting the inline function optimization etc. The optimization result of each optimization to the thesis algorithm is given behind each optimization and compares their difference. Finally listed result tables after various optimizations and displays the effects system diagram. One innovation point of this system is using all the optimizations based on the analysis of DSP operating mechanism to achieve the maximum optimization.The infrared video detection system discussed in the thesis makes it possible to achieve around90%of pedestrian identification detection rate in the whole identification process. The processing speed reaches20frames/second on the DM6437after optimization, reaching the video real time display. Therefore the system has strong engineering application value.