Design On Image Processing System For Defective Capsules

With the rapid development of the science and economy, people's living standards have improved significantly. At the same time, people request higher and higher testing requireme- nts for the drug quality that relate to their health . Buy capsules, we can usually judge by the naked eye whether is there damaged on the capsule, or medicines full of capsules. However, mass production of drugs in the pharmaceutical only one by one by the staff of the naked eye to determine the contour of drugs and dose, certainly can not meet the requirements of modern industrial production. Based on this background, this paper combined with the rapid development of the digital image processing technology in recent years,the core part of the capsule sorting system - hardware design, the damage capsule detection algorithm, system software design carried out relevant research, and do the simulation algorithm.On the hardware side, in order to meet the requirements of real time testing capsule, proposed to form the core of capsules image processing systems use DSP + FPGA, focusing on the TI's TMS320DM642 digital signal processing chip. And the CCD camera for the system,the video decoder chip TVP5150PBS, the video encoding chip SAA7121H, DSP's peripheral circuits and DSP connects External devices made a detailed introduction.On the capsule detection algorithm side, firstly introduces the common color model. Then introduced the two types of common noise in the hardware system. For these noise, several denoising methods are proposed. By MATLAB simulation, considering the time and in the premise of the treatment effect, select the median filter and Wiener filter as a capsule image preprocessing. Middle of the image processing stage compared several post-edge extraction algorithms to extract the capsule edge by the Soble edge algorithm. In the segmentation stage, after numerous tests, obtained when the threshold is set to 208 for post-segmentation better. In the final stages of the testing, from the concept of the image matching, extending the "image not matching" concept, through several experiments and found that the set the rate of not matching as 5% can detect the defective capsule. the purpose of detecting the defective can be achieved.On the software design side, introduces the development environment of DSP CCS3.3, system's debugging tools, COFF file and programming steps in detail.For C6000 series DSP program on the basic structure, DSP low-level driver made a detailed description.