Design And Implementation Of Multi-functional Industrial Camera System Based On ARM And Embedded Linux

Industrial camera is the core component of machine vision sensor system, which is generally used as a front-end in machine vision sensor system in order to complete image acquisition function. Traditional industrial cameras only have image capture and transmission functions. Image processing and analysis depend on the rear end of the PC, which causes massive system, high power consumption and high installation costs, so it is not suitable for independent work in the field. With the rapid development of computer technology, semiconductor technology, low power consumption and the features of the embedded CPU, such as, low cost and high performance, it appeared domination in the field of the complex calculation and the interactive applications. Therefore, based on embedded systems technology, multi-functional industrial camera which can work independently gradually become a hot topic in machine vision systems at home and abroad.For solving the shortcomings in the traditional industrial camera which has single function, strong reliance on PC, and can't work independently, a multi-functional industrial camera based on ARM and embedded Linux(hereinafter referred to as the camera) is proposed. In the aspect of hardware design, it first analyzes the system framework of the camera, compared with the several popular embedded solutions. By comparison with the embedded CPU, image sensors, image data caching scheme and communication interface design scheme, this scheme which select TI's AM3354 as master CPU of camera, MT9M001 as an image sensor chip, CPLD with SRAM as a FIFO buffer for image date, and USB, UART and Ethernet interface as the camera's external interface is the best choose. Finally based on the analysis of this program, we complete the circuit and PCB design of the camera.The thesis also analyzes the software architecture and the design of this camera in detail. In the operating system level, we complete the configuration and use of uboot, Linux kernel configuration and tailoring, Linux root file system production, camera driver development and so on. In the application level, the OpenCV library is uesd to provide a powerful image processing support. Meanwhile, in order to facilitate the secondary development, a SDK API interface is supposed to control the camera peripherals and interfaces. In the end, this design tests the video function of the camera, and uses OpenCV for contrast enhancement, histogram generation and edge detection algorithm testing, and respectively compares the algorithm running time on PC and camera. The test result shows that the design and implementation of multi-functional industrial camera can achieve 30 FPS frame rate when takes pictures of 1024x1280 resolution. And it also can use the built-in OpenCV processing and analysis of local image.Through the above work, the multi-function camera finally achieves the following characteristics: it has a wealth of communication and control interface on hardware, which allows data transfer easily and interactive control; it supports the flexible expansion of hardware and software, which can dynamically add hardware and software functions; it provides the SDK, which is convenient for secondary development quickly; it instals OpenCV library to provide a powerful image processing support.