Design Of Industry Smart Camera Based On Blackfin

China is the world’s factory and has numerous kinds of product, and their quality of all the products need to be detected. In many of the quality detection methods, machine vision technology has been applied widely in product quality testing field for acquiring the most abundant product quality information. Since CCD (Charge-coupled Device) and CMOS (Complementary Metal Oxide Semiconductor) image sensor technique and the performance and data communication speed of various microprocessors get greatly improvement, making it possible to integrate various functional modules into a single device, so there is the industrial smart camera technology.Industrial smart camera falls within machine vision technique, and it integrates image acquisition, processing and communication capabilities into a single camera, realizing work offline. Compared with traditional machine vision system relying on PC(Personal Computer), the industrial smart camera enjoys many advantages, such as multi-function, modularization, high reliability, and easy to realize, but it does not have as high operational speed as PC, moreover, in the current market once industrial smart cameras are out of factory, their functions are fixed, which makes it unable to meet user’s changing needs. Therefore, for the specific application requirements, this article designs an industrial smart camera, in which, Blackfin series DSP(Digital Signal Processing) processor BF531is used as the core of image processing unit and CMOS image sensor is applied to capture image data, and some removable units are added.This article focuses on the versatility of industrial smart cameras. In the hardware, for different application environments, image processing methods and control modes, a variety of configuration plans are proposed and the circuits of all modules are designed based on different hardware configurations. In the software, OpenCV(Open Source Computer Vision Library), a open source image functions vision library, is selected as a basic image processing library to design software modules, and different software modules are called according to the varying classes of the detected objects, which not only reduces the cost but also meets the diverse needs of users. Finally, the practicability of the software designs is verified by taking PCB(Printed Circuit Board) quality testing as an typical example.