The invention discloses a method for detecting macroscopic mura defects on a large-size
liquid crystal display panel. The method comprises the following steps of collecting a plurality of frame imagesof different areas on a
liquid crystal display panel to be detected according to a preset frequency and a
relative motion path, splicing all the collected frame images into a complete
panorama according to the physical coordinates of the frame images, iteratively compressing the complete
panorama for multiple times by using an
image compression algorithm to obtain a compressed image of the liquidcrystal display panel to be detected, carrying out compensation
processing on the compressed image of the to-be-detected
liquid crystal display panel to obtain a preprocessed image of the to-be-detected
liquid crystal display panel, carrying out the
contrast enhancement processing on the preprocessed image of the liquid
crystal panel to be detected by using a contrast-limited
adaptive histogram equalization algorithm, and carrying out binary
morphological processing on the enhanced pre-processed image to obtain an adaptive
binary image, the adaptive
binary image comprising a plurality of spots used for representing the macroscopic mura defect information, and obtaining the macroscopic mura defect information according to the spots.