Detection of the number of accessories in the milk powder cans in complex background

HE Li，LI Bo-quan

（School of Mechanical Engineering，Jiangsu University，Zhenjiang 212013，China）

Abstract：Aiming at detection requirements of the number of spoons and other accessories in the milk powder cans with the bottom of pock arched aluminum foil，a new adaptive morphological filter-based method for extracting foreground in the complex background was presented and a real-time detection system based on machine vision was designed. In the ball integral light source，the image was acquired，and it was processed through the image smoothing，template matching，thresholding，connected region，and adaptive morphological filter. After such processing，the quantity of milk spoon was detected by the statistic analysis of pixels of the extracted parts. The experiment results show that the rate of right detections is 99.99%，and the rate of false detections is less than 0.05% over the on-line detection. The designed system is suitable for detection of accessories，foreign material and defects of cans wall in the various milk powder cans or alien cans.