Binocular automatic particle inspection machine for bottled medical liquid examination

Automatic Particle Inspection Machine (APIM) is capable of detecting visible impurities in ampoules and penicillin bottles. However, for Large Volume Parenteral (LVP), Dynamic Background Interference and Narrow Depth of Field problems mostly lead to failure using traditional monocular inter-frame difference algorithm. In this paper a binocular inspection algorithm and system based on monocular machine is proposed to overcome those disturbances. The binocular inspection system only changes the pose of cameras without increasing hardware costs and the algorithm employs an efficient sequential-image-stereo matching method to achieve real-time online inspection. Experimental results of inspection comparison between monocular and binocular for alarm rate and omission rate show detection validity even if the blurred parts due to the disturbance cannot be distinguished from its own background by monocular method.