Design And Implementation On Vehicle Active Safety Systems With Panoramic Image

There are many blind sight areas around the vehicle, which will cause drivers frequently unable to get the road conditions. As a result, it will lead to traffic accidents. Paper builds panoramic image around the vehicle by machine vision to get rid of blind sight areas. Paper also designs an obstacle detection device system that can warn driver dangerous vehicles which are too close to it by sound alarm. In this way, they will improve active safety of the running vehicle, which is so important to us.Paper designs a vehicle active safety system with panoramic image. This system consists of four wide-angle cameras set on the vehicle body which are in four directions. They grab image around the vehicle. After that, image is transformed to top view of the vehicle. In the end, the four channels of image are stitched to a panoramic image from top view.Paper determines the image acquisition mode with four wide-angle cameras set on the vehicle body which are in four directions aiming at getting the surrounding video of the vehicle. Paper also determines the dual-core chip named TMS320DM6446 which is developed by TI company as the core chip. This chip has strong computing capability, high speed of image processing, good stability and convenience in vehicle. Paper determines the video decoding chip named TVP5158 to meet the need for processing 4 cameras image signal in the same time. The ARM and DSP embedded system solutions consist of TVP5158 and TMS320DM6446.Paper designs embedded Linux panoramic images of active safety systems by using ARM processor as the program manager to meet the system needs. Paper designs the system hardware circuits and builds the software development platform. The system hardware circuits consist of power supply protection circuits, the voltage conversion circuits, TVP5158 peripheral circuits and so on. Power supply protection circuits aim at providing stable power supply for the core plate and camera. The voltage conversion circuits are mainly used to provide different voltages for each core and I/O port on chip. TVP5158 peripheral circuits include a 4-channel video signal input and output circuits and chip peripheral pins voltage circuits. Software development is based on TI’s Da Vinci technology, with building compiler environment on Linux virtual machine system for DM6446. It consists of installing the Monta Vista kernel and DVSDK tool chains and setting cross-compiler environments. Software development process is mainly realized through NFS file system, which is anchored to, to avoid the frequently programming.Paper develops a feature-based obstacle detection algorithm and a panoramic image stitching algorithm based on coordinate transformation on the DM6446’s DSP side by using the designed embedded system platform. Paper verifies the hardware system able to collect and process the 4-channel video signal at the same time through actual test experiments. It meets the design requirements. Meanwhile, obstacle detection algorithm and panoramic image stitching algorithm are effective by actual function tests. At last, paper verifies the research results by whole real car experiments.