Sub-Pixel Defect Detection for Super Large LCD Images Using Quadrant Gradient Operator

This paper proposes an automatic sub-pixel component defect detection algorithm to locate the flaws of liquid crystal display (LCD) using super large images. To ensure detection accuracy, a line scan camera is used to capture the images of an LCD with more than 200 million pixels. Therefore, a tiny sub-pixel component of an LCD is mapped into many pixels in an image. The quadrant gradient operator is proposed to adjust gradient direction dynamically in different quadrants. Our method can directly detect sub-pixel component defects without extra operation to deal with edges. To detect defects efficiently and automatically, related parameters are estimated using a small random sampled sub-image. The projection method estimates the spaces between sub-pixel components, based on which the sample size and the defect area threshold can be estimated. Theoretical analysis and experimental results show that the proposed algorithm can accurately detect sub-pixel component defects in linear complexity O($n$).