Implementation Of Embedded Network Image Processing System

This paper studies the design and implementation of an embedded network image system based ARM microprocessor and FPGA. The whole system consists of three parts:image acquisition, image processing and image transmission.Samsung's S3C2440A ARM9 chip and Xilinx's Spartan-3E XC5S145E are the core hardware of the system and we use open-source Linux as an embedded operating system.As a front-end image acquisition module, FPGA is mainly to complete the image acquisition and preprocessing. Image data collected by image sensor controlled by FPGA will be binarization in accordance with a threshold value. The processed image data will be stored in the SRAM int the format of (X1,X2, Y).As the core of the entire image processing system, ARM is responsible for image processing and transmission. Image processing is to handle image data in a specific format stored in SRAM, consisted of MARKER detection and calculating of the two-dimensional coordinates of the binary image.After that, we send the image data to the computer through the LAN.When the observrd object moves, the trajectory of MARKER can be used to record the trajectory of the object. Recording a small amount of information of the MARKER points can obtain the movement of the observed object.The hardware design of the whole system includes the FPGA circuit design of image acquisition interface, ARM peripheral circuit design which incudes the circuit design of SDRAM, FLASH, USB, serial port, DM9000 network card and SPI,IIC bus.The software design of the whole system includes the migration of Linux, the device driver design and the application design. The migration of Linux includes the bootloader migration, linux kernerl and root file system migration. The device driver design includes the drivers for SRAM memory, S3C2440A's SPI bus and GPIO. The application software includes the design of the network transmission, the algorithm of MARKER detection and the design of calculating two-dimensional binary image centroid coordinates.The image acquisition, processing and transmission are integrated in an embedded target board by the system to send image data at 60 frames in a second.