Research On Technologies Of Vision-based Automatic Inspection Machine For Pharmaceutical Ampoule

Ampoule injection is one of the five preparations of the pharmaceutical industry in China, which is a wide range of clinical application, and it occupies a very important position in modern clinical medicine. Because of being affected by the production process, there will be many unqualified products for the ampoule injection as insoluble visible foreign particles and cosmetic defects. Traditional manual detecting can’t meets the requirement of automated production for high-speed and high-precision, and foreign inspection machine is expensive, high costs for maintenance, domestic inspection machine is immaturity. Therefore, vision-based automatic inspection machine for pharmaceutical ampoule that been researched in this paper has application value and economic benefits.Relying on the National High-Tech Research and Development Plan and National Natural Science Foundation of key projects, the article develops the research on the vision-based automatic inspection machine for pharmaceutical ampoule, the main work of this article as follows:1. The article elaborates industry background and the relevant state laws and regulations of ampoule injection, obtains the significance of research on the vision-based automatic inspection machine for pharmaceutical ampoule. And overviews of the researching status and development trend of vision-based Inspection machine for liquid pharmaceutical at domestic and foreign.2. The article designs different detection stations for visible foreign particles, cosmetic defects and capacity. Designs a distributed tracking program of intersecting-apart by bottle, the detection rate will reach to 600 bottles per minutes, and this distributed tracking program can guarantee the detection accuracy at the same time. Then, designs of the control systems of the optical and electrical integration, sucker sorting system, and optical imaging system etc.3. In view of the problem for deflection of the ampoule head in image, a fast rotating image method of bilinear interpolation-the reverse cumulative has been proposed; the article designs a method based on the edge features of the ampoule head to detect the cosmetic defects. And a fast image registration algorithm of improved NNPROD based on climbing searching to solve the deviation of the image sequences has been given; Based on the imaging conditions, the article designs of a classification extracting method for the trajectory of foreign particles to achieve trajectory extraction for different gray scale and shape.4. A distributed system is built that PC is responsible for image processing and PLC is responsible for controlling, and the article designs of detection software and control software that achieved to the aim of good human-computer interaction. Finally, develops vision-based pharmaceutical automatic inspection machine, and the machine has pass out the Knapp-Kushner test, results show that the machine can replace the manual inspection, the machine has been successfully applied to domestic pharmaceutical manufacturers.