Research And Implementation Of Embedded Intelligent Vision System Based On Blackfin

Different from the traditional vision system, the embedded intelligent vision system is embedded in the image sensor. It is a dedicated computer system which is designed for a special application. Compared with the traditional PC-based machine vision system the embedded intelligent vision system is more intelligent, more reliable, more integrated and more extensible. Therefore the system is more and more widely used in industrial, military, medical and many other fields. Now it has become a hot research field of machine vision.In this dissertation the research status of machine vision system and the trends of embedded intelligent vision system had been studied. Combined with the latest research results of this field, the new embedded intelligent vision system was developed based on the ADI’s Blackfin series DSP ADSP-BF60 x. The 60 fps image sensor MT9V02 x was used in this system to capture image data. Then the system could get the result of the image processing through ADSP-BF60 x processor without PC’s further intervention. The system could also send the measuring result, image data and command to the external system through the Ethernet, UART and general I/O interface. The embedded intelligent vision system of this dissertation abandoned the PC-based architecture reducing the complexity of the system. At the same time due to the use of high-speed DSP the speed and precision of image processing was greatly improved.The design and implementation of the embedded intelligent vision system’s hardware solution, software framework and application case were described in this dissertation. The comparison and selection of system solutions, and elaborated the design and implementation of each module were described in the hardware design part of the dissertation. The transplant of multi-tasking real-time operating system uC/OS-III and configuration of lightweight LwIP Ethernet protocol stack were introduced in the software framework part of the dissertation. The design and implementation of resource allocation and scheduling policies were alos described in this part. In the end of this paper the further research aspect was concluded by the analysis of the system’s function and performance through the result of optical character recognition application case.The ADI’s newest high-performace dual-core DSP ADSP-BF60 x and embedded real-time operating system uC/OS-III were used in the design of this dissertation. The embedded intelligent vision system combined hardware image processing and DSP software image processing to achieve high-speed digital image processing and complex multi-tasking This system could recognize paper currency number more than 24 frames per second with image resolution 752×480. It is a great embedded intelligent vision solution with high pratical value. To our knowledge, the architecture with ADSP-BF60 x and uC/OS-III for embedded intelligent vision product has not been reported in our country.