A Region-Scalable Fitting Model Algorithm Combining Gray Level Difference of Sub-image for AMOLED Defect Detection

摘要

In this paper, we improve the Region-Scalable Fitting (RSF) model by the gray level difference of sub-image for RSF model's sensitivity to initial contour and slow speed in active matrix organic light emitting diode (AMOLED) defect detection. The gray level difference of sub-image algorithm can only locate the approximate defects area. The Region-Scalable Fitting (RSF) model overcomes the detection difficulty caused by the intensity inhomogeneity, but the local characteristic makes it extremely sensitive to the position of the initial contour curve. To solve this problem, we combine the gray level difference of sub-image algorithm with the RSF model. Firstly, the approximate position of the defects area is found by the gray level difference of sub-image algorithm, and the outline of this approximate defects area is taken as the initial contour curve of the RSF model. Then we implement the RSF model to segment the defects accurately. The experimental results show that the use of gray level difference of sub-image algorithm to locate the initial contour overcomes the disadvantage that the RSF model is sensitive to the initial contour, and improves the detection speed.