Detection instruments of flat panel displays mainly include visual inspection instrument, array detection instrument and so on. With the increasing evolving of production line’s generation of flat panel displays, which from 6G,7.5G,8G,to8.5G, the size of the plate assembly is increasingly greater and corresponding detected visual field is also growing. Because of continuous improvement on defect resolution, it’s unable to expand the visual field by the method of reducing imaging lens’ magnification. This paper relies on the major scientific equipment developed special projects "Development and Application of Flat Panel Displays’ Automatic Optical Inspection Instrument" and applies in TFT-LCD precision assembly production line for the requirements of high-precision imaging at wide range to achieve high-speed and high-resolution scan detection. This paper mainly studies on cameras selection, lens design, design of cameras adjustment mechanisms, lighting system design and image stitching technology.In TFT precision assembly production line, images of the measured object can be obtained by visual inspection in using of high-resolution and high-sensitivity imaging techniques and the size, position, direction and defects information of the target from images can be obtained through the fast image processing and pattern recognition algorithms so that can achieve target products’ recognition and identification, targeting and guidance of assembly agencies and such tasks in line of inspection and assembly. To solve the high-precision measurements in large field, this topic selectsl6k series of line scan camera to scan imaging and designs a lensM72, which used to adjust focal length. In order to achieve high-precision imaging, a 5 DOF installation adjustment mechanism is designed. LED cold light source with fiber-optic bundle guide is selected to consist of lighting system and high contrast images can be obtained by adjusting camera’s imaging angle and the angle of light source. Image stitching is the difficulty in scanning imaging, two points of minimum error which obtained in x-direction and y-direction using the least squares method are considered as reference points for splicing image, which improves the accuracy of image stitching.Experimental results indicate that high-contrast images can be obtained when the angle of light source varies from 17 degrees to 45 degrees and the angle of the camera varies from 0 degrees to 17 degrees. Higher precision stitching effect can be achieved when randomly selecting 10 points based on the least squares method. |