Research And Implementation Of Embedded Intelligent Industrial Camera

Traditional machine vision systems are separated from industrial cameras and PC in the structure.This traditional machine vision system relies on PC for image analysis and processing,which has many disadvantages,such as high power consumption,high cost,large volume and limited installation in special environment.In order to meet the needs of miniaturization and portability of users,industrial cameras are developing towards the direction of high integration,high resolution and intelligence.Intelligent industrial cameras,its essence is a highly integrated micro machine vision system,which has been widely used in the field of semiconductor,precision electronics industry and quality inspection etc..This paper combines embedded technology and image processing technology,designs the hardware and software of intelligent industrial camera,realizes the embedded intelligent industrial camera which integrates image acquisition,image processing,communication and other functions.This paper mainly studies from the following aspects:Firstly,we analyzed the framework of intelligent industrial camera.Several system development schemes are proposed from three aspects: processor,image sensor and image data transmission interface.We choosed the ARM processor S5P6818 which based on Cortex-A53 architecture,OV5640 and MT9M001 CMOS image sensor as core components,alse adopted DVP interface as the solution of image data transmission.The image sensor module,power supply circuit,serial port,USB and Ethernet interface circuit are designed in this part.The schematic diagram and PCB wiring diagram are drawn.The hardware design and manufacture of intelligent industrial camera are completed.Then we introduced the basic structure of embedded Linux system and the methods of building intelligent industrial camera development environment,and studied the contents of BootLoader,Linux kernel and root file system.On S5P6818 development platform,the driver of CMOS image sensor OV5640 and MT9M001 are designed and developmented,which based on V4L2 driver framework and DVP interface.Finally,we designed the program of local image acquisition by calling API of V4L2,and transplanted OpenCV to the camera.Realized the local license plate image processing by the camera itself,extracted the license plate area,and completed the application software development of the intelligent industrial camera.Through the above work,the ultimate intelligent industrial camera can tailor the hardware and software of the camera according to the application needs,which enhances flexibility and saves costs.Local image acquisition,processing and analysis can be realized on the camera,and the integration of machine vision system can be realized.Compared with general intelligent industrial cameras,it also has the advantages of supporting two types of camera switching,multiple types of communication interfaces and secondary development.