Research On High Throughput Automatic Colony Picking System Based On Machine Vision

Bacterial selection procedures in gene sequencing and microbial synthesis experiments in the field of biotechnology require researchers to select and select the cultivated microorganisms because of their small size,large number,high repeatability,heavy workload and long time consuming.In response to these problems,developed countries such as Britain and the United States took the lead in research and developed different selection and cloning systems to replace manual operation and realize automatic operation,so as to reduce experimental time,improve the accuracy of selection and improve the efficiency of bacteria selection.However,most of the cloning systems are single or eight channel needle picking,and the size of the cloning equipment is huge.Some domestic research departments are mainly related to the improvement of imported equipment,high cost,closed technology and single function,low market share.Combined with the actual needs,this subject designed a high-throughput automatic colony picking system based on machine vision.The main research work is as follows:(1)The overall scheme of automatic colony picking system is designed,including mechanism design and hardware selection.The implementation mechanism is designed to be 32 channels.Selecting metal needle,adding cleaning and disinfection module,avoiding the time consumption caused by needle change operation,and can be recycled;The overall structure layout is reasonable,the operation space is compact,and the operation time is reduced on the basis of meeting the requirements of function and precision.(2)Colony image was collected based on machine vision and colony preprocessing was carried out.A feature point extraction method combining concave point detection and centroid extraction was proposed to segment the adhesive colonies,which could solve the common series,parallel and complex adhesive colonies.Colonies were identified and counted based on flood filling algorithm,and key information such as centroid coordinates and radius of colonies were obtained.(3)A selection model was established based on the TSP model and the motion control process,and the improved simulated annealing algorithm(I-SA)was used to optimize the colony selection path.In the improved algorithm,the path initialization module adopts a new path initialization allocation method,the path generation module adopts the "rough exploration"and "precise exploration" modes,and adds the tempering function in the temperature drop module.The simulation test of the algorithm shows that the I-SA algorithm is better than other path optimization algorithms.The actual selection experiment shows that the selection efficiency of the I-SA algorithm is 14%-19%higher than that before the improvement,which indicates the feasibility and effectiveness of the algorithm.(4)Based on the TSP model and the motion control process,a selection model was established,and an improved simulated annealing algorithm(I-SA)was proposed to optimize the colony selection path.In the improved algorithm,the path initialization module adopts a new path initialization allocation method,the path generation module adopts the "rough exploration" and"precise exploration" modes,and adds the tempering function in the temperature drop module.The simulation test of the algorithm shows that the I-SA algorithm is better than other path optimization algorithms.The actual selection experiment shows that the selection efficiency of the I-SA algorithm is 14%-19%higher than that before the improvement,which indicates the feasibility and effectiveness of the algorithm.