Font Size: a A A

Design And Realization Of Recognition And Location System In High-Speed And High-Precision LED Die Attach

Posted on:2018-03-12Degree:MasterType:Thesis
Country:ChinaCandidate:H X LiuFull Text:PDF
GTID:2348330542970387Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
With the support of the policy,LED industry of China has been leap in the development,Industry practitioners up to 200,000 people,more than 6,000 enterprises,forming a covering substrate,epitaxial,chip,package four aspects of a complete industrial chain.As the key equipment of LED packaging,Die-Bonder is in great demand.Currently on the market of domestic Die-Bonder performance is generally poor,imported Die-Bonder performance is high but expensive.LED industry is an urgent need for cost-effective domestic Die-Bonder appears.Around the entrusted enterprises to the core of Die-Bonder to take die attach target recognition and positioning technology development needs,this paper carried out the research work,the specific contents and achievements are as follows:(1)Based on the analysis of the functional requirements and performance indexes of high performance Die-Bonder,has chosen the industrial camera,lens and light source as the core components,and built the visual experiment system of Die-Bonder.(2)Proposed to extract the characteristics of the weak classifier and continue to upgrade to a strong classifier to achieve die attach target--crystal and cup identification of the positioning of ideas.The three characteristics of Haar-like,HOG and LBP of die attach target are studied,extracting single feature to construct weak classifier,then,combined with Discrete AdaBoosting,Real AdaBoosting,Gentle AdaBoosting,LogitBoosting four classifier lifting algorithm to get 12 cascade classifier of the die attach target.The experimental results show that the Haar-like + Discrete AdaBoosting cascade classifier works best.A die attach target recognition and positioning algorithm is designed with above classifier as the core,and Verified to meet the needs of high-speed high-precision solid crystal machine Die-Bonder.Compared with the commonly used gray-scale template matching recognition and localization algorithm,the positioning accuracy of the cascade classifier identification and positioning algorithm proposed in this paper not lower than the gray pattern matching recognition and localization algorithm,and the positioning speed is much higher than that of the latter.(3)The other two factors that affect the performance of solid crystal machine are taken as follows:Take crystalstrategy and control logic are studied,an improved sampling strategy is proposed,and a feasible control timing is designed.At the same time,the visual software of Die-Bonder machine is designed and realized.The application shows that the solid crystal vision software based on the cascade classifier identification algorithm can meet the need of taking solid crystal.
Keywords/Search Tags:LED, Identify positioning, Die-Bonder, Cascade classifier, Haar-like
PDF Full Text Request
Related items