Pearl Quality Detection System Based On Image Processing

Pearl is one of the most important aquatic products. Pearl plays an important role in the aquaculture of our country. Pearl industry has a long history in China, and it is one of the most productive parts of Zhejiang's industry. China is the earliest country who discovered pearls. The production of China's pearls is the largest in the world. However, in the process of producing, pearls are selected manually by people. It is hard and tiring work. No automatic system was applied. So designing an automatic pearls selecting system can greatly improve the development of pearls industry.In order to classify pearls, we have to define the class of pearls according to the shape, color and flaw of pearls. In our system, we use three cameras to take photos of pearl from three different angles. By analyzing these photos, we could get the information of pearls' shape, color and flaw.In the system, we use a global light as the light source. So the pearls have even lights shined on their surface. By analyzing photos of different angles, the system obtained the information of pearls' color, whorl, shape and texture. According to the information, the system can classify the pearls by controlling the IO card. And we use infrared ray as trigger, so the system could automatically take photos and process the photos.In this paper, I try to introduce the realization of the system, including hardware controlling and parts of software. In the software part, this paper tries to present the color recognizing module and whorl detection module. Support Vector Machine is used as color classifier. With the help of SVM, the system can recognize red, white, yellow and purple pearls. In order to detect long flaw, we use technology of bounding box and direction checking. The paper also shows how to detect non-consecutive whorls.