Research And Implementation On Counting System For Quasi-circular Objects

The image segmentation and recognition is one of the most important research directionin the field of computer vision and image processing, particularly the image of Quasi-circular objects.Quasi-circular objects have properties of convex sets and high degree of circularity. They exist in every corner of our lives, such as medical cells,go pieces,pills, grain particles, the end surface of bundled steel bars and so on.Human life and productionoften need to count various types of quasi-circular objects, and the early work is mainly done by people.The long time in statistical process crops would cause extreme eye fatigue. As a result, the accuracy and counting speed are greatly affected by human factors. With the social development, the use of image processing technology quasi-circular objects counting more and more applications.But most of them only propose appropriate processing algorithm is directed to a specific quasi-circular objects.In addition, the quasi-circular processing algorithms have some shortcomings, such as: quasi-circular processing algorithms based on watershed segmentation will lead to over-segmentation class round object.In order to compensate for the lack of manual counting, we urgently need to develop a system which can automatically measure the number of quasi-circular objects, more importantly, the system improve quasi-circular objects statistical accuracy and labor productivity.An automatic counting system for quasi-circle objects under complicated working conditions is developed in this paper, based on image processing. The major aspects of this paper isfollowing:(1) Introducing common image processing technology to count thequasi-circle objects. Raising improved counting method after analyzing advantages and disadvantages of existing counting algorithm. Improved K-means clustering segmentation algorithm is utilized. By specifying the cluster centers in the YCrCb color space method effectively improved the quasi-circle objects segmentation, and greatly improves the system speed.Taking into account the quasi-circle objects surrounding environment is complex situations, the algorithm based on space vector to the complex is utilized. The algorithm improves the accuracy of the count.(2) An automatic counting system’s general structure is designed, and specific for grain particles, pills and go pieces are put forward. The system includescamera, image acquisition,image processing, results showing.(3) The prototype system, which based on VC and OpenCV, is built and images of the quasi-circle objects are recognized by presenting the counting way for quasi-circle objects in this paper. Experimental data show that the automatic counting system can count the number of quasi-circle objects accurately and work well.Quasi-circle objects counting system object of this paper can be quickly and accurately quasi-circle objects complex adhesions separated, effective count quasi-circle objects. The entire software system processing speed, high accuracy, with good real-time performance and adaptability.