Research On Image Matching Algorithm Based On Geometric Primitive Feature And Its Application In Solder Paste Printer

Solder paste printing is the first step in the SMT production line,Paste printing quality will directly affect the follow-up steps and the quality of products,In order to obtain good printing result in the current production condition,we mainly use visual technology to achieve printing accuracy.Image matching algorithm is one of the core elements of visual positioning technology,it plays an important role in the SMT production line.So it's meaningful to increase the positioning accuracy by improving the image matching algorithm.In this study,in order to increase the printing accuracy and take into account the printing efficiency of solder paste printer,we take GT type solder paste printer as the research object,we proposed a new matching positioning algorithm for the alignment markers,we established a visual system alignment model,at the same time,we analyzed the deviation of the alignment model.(1)We proposed a new geometric hash matching algorithm,which was based on linear primitives,for the mark-hole positioning problem of the circuit board and the SMT stencil in the process of Solder Paste Printer alignment.We constructed base coordinate system by using the relative location of the primitives.Then we established the primitive error zone and template hash table.Follow the steps above,the target primitive was projected into the template hash table,at last,the matching results can be obtained by comparing the voting result.(2)We constructed the alignment model of the solder paste printer.We established four coordinate systems which were used for representing the positional relationship among the different parts.We established the transformation relationship between the coordinates.Then conducted key parameters calibration.We analyzed the rotary motion and translational motion of the platform and derived kinematic formulas for the platform,and eventually got the moving distance of each motor.(3)When we built the visual system alignment model based on the solder paste printer,we conducted error analysis of the alignment model,the analysis is composed of mechanical structure error analysis and pixel error analysis.(4)We tested the speed and accuracy of the geometric hashing matching algorithm which is based on linear primitive,the test was conducted on the solder paste printer which is GT type in Dongguan GKG company.The test results showed that the speed of the matching algorithm we proposed is faster than that of the traditional matching algorithm.at the same time,the accuracy of the matching algorithm we proposed is higher than that of the traditional matching algorithm.So it has important application value for improving the accuracy of solder paste printer.