2017 年 83 巻 3 号 p. 258-262
A novel image quality inspection method that detects low contrast “mura” (non-uniform regions) has been developed by analyzing the Fourier spectrum of the images. The inspection images are transformed to Fourier frequency space, and the power spectrum is filtered using CSF(Contrast Sensitivity Function) to obtain the human eye sensitivity. The radial frequency and angular frequency histograms of the power spectrum are used to detect the image defects. A “mura” model experiments were executed to verify the novel method, and the experiments using 62 real samples showed good detection ability for the “mura” defect samples.