Detection of foreign substances mixed in a plastic bottle of medicinal solution using real-time video image processing

This paper presents a new method for detecting foreign substances mixed in a plastic bottle of medicinal solution using real-time video image processing. Obstructions to detection of foreign substances due to unevenness such as corners, embossed symbols and graduations on the bottle surface are removed by looking through a clear part of the surface using an optics obliquely imaging a vertical section passing through the center of the bottle. 3D scanning of a solution to detect foreign substances is performed by rotating a solution and making it go through the vertical section, which is imaged on a CCD image sensor device successively at video frame rate. A foreign substance can be detected as a black spot in diffusing background illumination or as a luminous spot in the illumination of collimated light by using appropriate illuminating and imaging optics. Black spots or luminous spots are detected by comparing values of corresponding pixels of two successive frames and by selecting a smaller or larger value. Traces of foreign substances are obtained after minimum or maximum value selection is done for all video frames covering a bottle of solution. A detection rate of 90% at least was experimentally achieved for substances larger than 0.3 mm in diameter such as pieces of carbon rubber, human hair styrene resin, and chemical fiber Detection time was a second for a 1000-ml bottle