Research Of Capsule Defect Detection And Quality Evaluation

With the quality of medicine is paid more and more attention,people also have more research on the quality of capsule.But,there are some factors influencing the quality of capsule detection algorithm,like the quality of image is not good,the detection type is single,or the algorithm is complex,the speed is low.Based on the analysis of capsule defect,the paper proposes a set of capsule defect detection and quality evaluation algorithm.The algorithm is using the information of image segmentation,and basing on the gradient information to seek spots.The main work of this paper as follow:First and foremost,the capsule detection is based on capsule localization.Aim at the effects of image,like the light,the paper locate the capsule with the knowledge of digital image processing.It correct by the gradient information of the capsule boundary,and the angle,to get the fine profile of the capsule.Secondly,basing on the information of the standard capsule contour,the algorithm access to the total length of capsule.And it distinguish and testing the capsule length,including the total length of capsule,the length of capsule cap,the length of capsule body,Through getting the information of capsule nested portion,it follows by using the the gradient level projection.Furthermore,it detects the spots as the whole capsule image and the nesting part of capsule.Through the spot portion of higher gradient to obtain its regional position.Then,it use the mean filter to remove noise and detect capsule spot-The detection of nested portion is getting its region and for this part of the separate spot detection.Finally,it use the above algorithm in a small sample database of NEU-CID library,for defect detection,of which have 277 capsule image,a total of 831 capsule,the type of defect detection accuracy reached more than 96.5%.The effect is very good.