Research On Key Image Processing Technology Based On DMD Lithography Machine

As one of the core technologies in the field of semiconductor processing,photolithography has the characteristics of high precision and low cost.In recent years,maskless lithography machines have been rapidly developed.With the development of digital micromirror devices(DMDs),DMD-based digital light processing(DLP)has also been widely used in maskless exposure machines due to its flexibility,parallelism,and high speed.The application plays an important role in miniaturization and high-precision lithography equipment.Accuracy is an important objective pursued by the lithography equipment.The step-by-step projection exposure based on the DMD is highly efficient.However,since the edges of the image need to be stitched several times,the inaccuracy of the stitching needs to be handled.In addition,the projection lens may cause nonlinear distortion due to process or assembly.Therefore,the lens must be corrected for distortion to improve the accuracy of the system.In this paper,the related digital image processing technology is studied based on the image stitching error and lens distortion in the exposure process of the DMD lithography machine.By analyzing the specific mode of the step-exposure mode,it is clear that graphics splicing is the main technical difficulty of this mode,and the causes and common situations of the image splicing error are analyzed.The DMD pulse gradation modulation technology is introduced to achieve different gray values of pixel values.level.In order to solve the situation of misalignment and overlap generated during the stitching,it is proposed to increase the overlapping area for each adjacent frame image,and use the sine function as the grayscale variation function of the overlapping area,so that the overlapping area and the normal area can be well overexposed.The goal of seamless splicing.At the same time,using grey modulation technology to weaken the sudden mutation caused by stitching misalignment,the dislocation can be smoothly transitioned and verified by simulation.In order to improve the accuracy of distortion correction,a new method of projection calibration is proposed.Firstly,the optical axis of the projection system is determined to be perpendicular to the projection plane.The checkerboard calibration plate and the projection image are in the same plane.After that,the camera is calibrated using a conventional camera calibration technique.The external parameters of the camera have obtained the conversion relationship between the camera pixel image and the projected image,so that the projected original image without the perspective of the camera can be obtained,so as to reduce the projection error caused by the perspective error of the camera,and then to restore After the projection chessboard picture is used for edge extraction and edge curve fitting,and according to the Brown distortion model,the distance of each point on the distortion curve is increased by a weight when calculating the fitting distance deviation,and the farther the distance is from the center point,the greater the weight is.Small,in order to construct the deviation distance objective function,and in addition,considering only the radial distortion,according to the collinear principle of distortion center point,distortion curve point and fitting straight line ideal point,another constraint condition is constructed.Then use the simulated annealing algorithm to perform multiple iterations to optimize the distortion parameters.Finally,the accuracy of the algorithm is verified by simulation results.At the end of the article,an experimental platform is built in combination with the actual project.The camera takes the picture of the calibration plate and the parameters of the camera to restore the projected image.Then the distortion correction algorithm based on the straight line features proposed in this paper is used to correct the distortion of the restored projected image.Parameters,complete the calibration,so that the distortion correction error is reduced to 0.2pixel.By comparing with other methods,the high accuracy of this method can be demonstrated,which verifies the effectiveness of the proposed distortion correction algorithm.