Research Of Scallops Selection And Classification System Based On Machine Vision

Our country is a maritime power,the development of mariculture plays an important role in agricultural production and economic development. Selection and grading of scallops is an important part of the scallop acquisition procedure, valid and reliable selection determines the efficiency of the scallop culture to some extent.At present, selection and grading of scallops uses artificial way and it is not only low efficient and high strength, but also the results of the classification is influenced by artificial subjective factors. Mechanical sieving grading accuracy is low, collisions and vibrations often leads scallop to edge breakage, resulting in sick or dead one, so it has not yet been widely applied.This paper dedicated to the study of scallop selection and grading systems based on machine vision, machine vision inspection technology has the advantage of non-contact, high precision and high speed. Function of scallop grading system is the realization of automatic measurement and grading of scallop and it includes two parts:scallops detection (count) and scallops measurement. Some issues appear during the research, such as segmentation of different color surface, eliminate noise of water and dirt scallop image, fast and accurate measurement of scallop size, phenomenon of overlapping and side by side scallop. Multi resolution analysis help speed up the detection efficiency, Frame difference and median filter method help eliminate noise, and Combine with Otsu method and YCRCB color space to segment scallop images, this paper presented a size measurement method base on scallop convexity detects and complete the detection and measurement of scallops. Using the secondary treatment of local edge to further enhance the detection precision, analysis and treatment of overlapping and side-by-side problems to reduce the undetected rate.In order to verify the legitimacy and effectiveness of algorithm designed in this paper, we develop the Intelligent detection system of scallop based on the algorithm designed in this paper, and make test and application in a live environment. Results show that algorithm and system designed in this paper is robust on condition that ensuring stable and accurate measurement.