Research On Solder Joint Quality Inspection Method Based On Image Processing

With the rapid development of modern electronics industry,printed circuit board(Printed Circuit Board)has gradually appeared diversified,miniaturized,high-density technology products,and at the same time put forward higher requirements for solder joint quality inspection.During the production process,due to processing factors such as materials and processes,SMT products have various defects,such as: too little solder,too much solder,not soldered,etc.In the production line,the method of manual visual inspection is usually used to detect the product,but this method is time-consuming and labor-intensive,and the detection accuracy is far from meeting the needs of modern production.Research on the use of computer vision to replace the artificial vision for solder joint detection method to improve solder joint It is extremely important to check the recognition rate to ensure the reliability and stability of product quality.In view of the characteristics of the solder joint image of electronic components on the production line,this paper studies the solder joint quality detection method based on image processing to identify SMT solder joint defects in order to improve product quality and reduce production costs.The main work of this article is as follows:First,study commonly used image preprocessing algorithms,including image normalization,image filtering,image contrast enhancement,etc.,based on the preprocessing,extract the geometric features and wavelet features of the solder joints,and complete in the MATLAB experimental environment.Simulation experiments of different algorithms.Secondly,in order to simplify the problem,reduce data redundancy,and obtain more scientific and effective feature information,a feature fusion method based on principal component analysis is proposed to reduce the complexity of the experiment and obtain effective solder joint image features for subsequent solder joints.Quality testing has laid the foundation.Then,in view of the shortcomings of the BP neural network,such as long training time and easy to fall into the local minimum,a BP neural network algorithm based on improved particle swarm is proposed.Aiming at the problem that the deviation of some input weights and hidden layer existing in the traditional extreme learning machine algorithm is 0,which leads to the failure of the hidden layer nodes,a PSO-ELM solder joint classification model is proposed.Finally,an optimized BP neural network and an extreme learning machine model were established for experiments.The experimental results of the two different models were analyzed,and it was found that the detection rate of both was affected by the number of training samples,and the recognition rate of BP neural network was slightly higher Extreme learning machine.The experimental results show that the principal component feature fusion method proposed in this paper can improve the speed of solder joint detection,and the algorithm after particle swarm optimization has better quality inspection results,and improves the accuracy and efficiency of solder joint quality inspection...