I found a solution from Toshiba to recommend to you. Toshiba's TMPV7502 processor is needed here. The basic structural framework is shown in the figure below.
The system consists of two main parts: a rear-view camera, including processing hardware; and a display unit that enables designers to evaluate and test the system. The camera part of the reference system is based on four folded flexible circuit boards, which occupy a small space. The reference design is compatible with CMOS sensors from different manufacturers, leaving room for designers to specify sensor modules that meet the exact needs of the application. The imager is mounted on the first board, together with the associated EEPROM. The second board contains the main processor (TMPv7502) and the necessary working memory. The third board contains flash memory, as well as DIP switches for defining boot options and a manual reset switch to help during system debugging. The final board provides communication with the receiver board via the FPD-Link III interface and the HSD connector. The board also provides a micro USB interface, which allows direct connection to the camera system for software development, monitoring and debugging. Toshiba has developed, since first releasing its original image recognition processor in 2004, a dedicated product lineup. This includes the TMPV7502 in the 11mmx11mm LFBGA324 package, which is primarily targeted at ADAS applications where available space is constrained. The TMPV7502 is a RISC-based triple-core processor with onboard RAM and a total of five hardware-based accelerators for high-performance object recognition processing and low current consumption. The device combines high performance with low power consumption of <1W through a highly parallel approach – rather than serial processing with high clock frequencies. With an onboard CAN bus interface, the TMPV7502 is implemented in single-controller systems within rear camera modules. In the future, the most basic systems will have to be implemented by simply switching the video from the rear-view camera to the in-cabin screen – in just two seconds when reverse gear is selected. Many current rear-view camera implementations are already more complex than this and include a processing module that overlays trajectory lines on the screen for maneuvering and indicating obstacles, giving the user information for obstacle detection. The current challenge is to improve object recognition through image-based processing and include it inside the camera module. The new module needs to be the same size as the current camera module and of course introduce sealing to prevent water. This greatly increases the requirements for system suppliers, as a balance must be struck between processing power and dissipation losses. It must be taken into account that CMOS image sensors typically show increased noise performance at temperatures greater than 85°C.
| 
提升卡
变色卡
千斤顶
京公网安备 11010802033920号
