The Research On Video Capture And Edge Detection System Based On CMOS

CMOS video capture and edge detection system consists of CMOS image sensor, USB2.0 data acquisition system and PC displaying&processing software. It has many advantages, such as convenience for application, simple driving circuit, high transmmition speed and great density of integration. As a result, it's widely used in robot vision, character recognition, medical image processing and object extraction in remote sensing images.This paper has designed a real-time video capture and edge detection system based on color CMOS sensor by using USB2.0 interface technology as well as OpenCV computer vision library and Visual C++ 6.0 programming platform. Research and development work involved in this paper can be concluded as the following aspects:(1) Through studing the reference technique datasheet of color CMOS sensor OV7640 and sufficient analysis in synchronize timing signals, this paper has proposed and designed a scheme interfacing with USB2.0 chip. This interface scheme not only has highly expandability and reusability but also has the advantage of simple connection.(2) Accoding to the SCCB bus standard of Omnivision image sensor, an interface module which communicates with USB2.0 I2C bus has been researched and designed. By sending definite command through this module, the register value of CMOS sensor could be changed dynamically.(3) By taking USB2.0 chip CY7C68013 as the core, this paper has designed a high speed video data acquisition system based on Slave FIFO mode. This data acquisition system can reach the speed of 18.3MB/S, it can fully satisfy the demand of CMOS video system.(4) This paper introduces the basic main frame of Cypress CYAPI function library as well as its calling method and driver development. By applying Visual C++6.0, a real-time video displaying software has been programmed. This software can restore the USB2.0 video data to its original color format. More over by using its controlling module provided by the software, user can dynamically set the parameters of the CMOS sensor and control its working status. (5) This paper has introduced a multi-scale edge tracking approach based on mathematical morphology, while eliminating the noise this algorithm can detect the image edge efficiently. Based on OpenCV computer vision library, by using Visual C++ 6.0 as the programming platform, video image edge detection software has been compiled.