Study On An Industry Smart Camera Based On UClinux

Industry smart camera is an embedded vision system, with the abilities of image capturing, processing and data transmission. The advantages of modularization, miniaturization, and fast application development, make it become an important trend of industrial vision system. This dissertation designs and develops an industry smart camera based on u Clinux. The key techniques and implementations are discussed in detail, including hardware architecture, embedded operating system porting, embedded application development, application developing environment, and image processing algorithms. The main contributions are as follows:A hardware platform based on Blackfin-DSP and CMOS for the proposed industry smart camera is developed. The main circuit designs including processor, CMOS, data memory, communication, and external interface module are provided.Embedded system development technique for camera firmware is studied, cross compiler environment configuration and Uboot bootloader porting are discussed, customization and porting of u Clinux kernel are implemented. Furthermore, hardware driver and application software are developed. Hardware drivers for several modules, such as CMOS, FLASH, light control, external interrupt and so on, are implemented. IO multiplexing model and multi thread technology are utilized for embedded application development. The application layer is designed to provide the following functionalities: task management, image processing, Ethernet and serial communication.Graphical application development environment of smart camera is designed based on architecture of modularized application programming tool. Without writing any codes, user applications can be quickly built just by configuring the application programming tools. Implementation of the main function modules, including project management, programming tool, equipment management and communication protocol, are proposed.Aiming at the requirement of application programming tools, the image processing algorithm and implementation are discussed. Image preprocessing algorithms including filtering, threshold segmentation and edge detection, are analyzed. Target locator algorithms based on template matching, Hu moment, SITF are investigated. The fitting of circle, straight line, and rectangle is achieved using the least square method and contour approximation.Prototype of hardware and software coding for smart camera are achieved. Basic performance tests including image acquisition and maximum frame rate with different ROI are carried out, and also, the applications tests including geometric dimensioning measurement and optical mark identification had been carried out. Experiment results show that the system framework and implementation are effective and feasible.