Application Of Registration Optimization Algorithm In Character Registration

Due to the jitter of the acquisition system and the offset of the workpiece position during the process of acquiring the character image by the industrial camera,there is some translation,scaling and rotation of the actually obtained character image.In order to facilitate further processing and analysis of character images such as character defect detection,it is necessary to map the actually obtained character images into the standard character image coordinate system.Therefore,image registration techniques are used to complete the registration of the actual character image with the standard character image.In order to improve the accuracy and efficiency of the character image registration,the binary image character is used as the feature for the actual image registration.The number of different pixels in the reference image and the floating image is used as the registration optimization function.In order to obtain binarized characters accurately,the character images are filtered and enhanced for preprocessing,and then the character image binarization is performed using an image segmentation method.The particle swarm algorithm and artificial bee colony algorithm are used to register the character images and the improved algorithms are proposed respectively for the disadvantages of these two optimization algorithms and applied in character image registration.The content of this paper is as follows:The translation,scaling and rotation of the actual captured character image are not conducive to subsequent character image analysis such as defect detection.The use of image registration technology is proposed to improve the accuracy of character image analysis.Analyze the feature space,search space,similarity measure and search strategy of image registration.Combining with the characteristics of actual character images,this paper proposes an optimization function that uses the minimum number of pixel points of binarized character images as registration,and uses an intelligent optimization algorithm to improve the accuracy of registration.Aiming at the problem that the actual character image has noise and no obvious contrast,a character image preprocessing method is proposed to guide the filtered image and then use wavelet enhancement.In order to improve the wavelet enhancement effect,a piecewise nonlinear high-frequency enhancement algorithm is proposed to keep the edge information of the character image well,which is beneficial to the character image segmentation.In character image segmentation,an image segmentation algorithm based on multi-factor complexity is proposed to improve segmentation accuracy.For standard particle swarm algorithm,there is a problem of slow convergence speed and easy to fall into local extremum in character registration.An improved particle swarm algorithm is proposed,which adopts non-linear changes of inertia weight and accelerating factor,and improves particle search strategy to improve registration accuracy and convergence speed.Aiming at the problem that the standard artificial bee colony algorithm has poor individual exploration ability and slow convergence rate,an improved artificial bee colony algorithm is proposed.The non-roulette probability model replaces the original probabilistic model of following bee roulette,and the following bee search strategy using globally optimal individuals to guide,speed up the search speed,registration accuracy and convergence speed have improved.