| Title: |
Non-uniform Image Defect Inspection Method |
| Document Number: |
20080107328 |
| Publication Date: |
May 8, 2008 |
| Appl. No: |
11/933511 |
| Application Filed: |
November 01, 2007 |
| Abstract: |
A non-uniform image defect inspection method includes steps of inputting an original two-dimensional image; separating a non-uniform background image from the original two-dimensional image by Discrete Cosine Transform (DCT) to obtain a residual image without the non-uniform background image; binarization segmenting the residual image to extract defects from the residual image, wherein the segmented defects are the inspection results. |
| Inventors: |
Chen, Liang-Chia (Jhonehe City, TW); Kuo, Chia-Cheng (Yanshuei Township, TW) |
| Claim: |
1. A non-uniform image defect inspection method, comprising steps of: a) inputting an original two-dimensional image; b) separating a non-uniform background image from the original two-dimensional image by Discrete Cosine Transform (DCT) to obtain a residual image without the non-uniform background image; and c) binarization segmenting the residual image to extract defects from the residual image so as to obtain inspection results. |
| Claim: |
2. The non-uniform image defect inspection method of claim 1, wherein the step a) further comprise the follows steps: b1) performing a 2-D (two-dimensional) DCT operation on the original two-dimensional image to obtain DCT coefficients of a frequency domain; b2) selecting a cut-off frequency from the frequency domain and eliminating high-frequency DCT coefficients higher than the cut-off frequency; b3) performing a 2-D Inverse DCT (IDCT) operation to reconstruct a background image with a brightness distribution similar to the brightness distribution of the original two-dimensional image; and b4) subtracting the background image from the original two-dimensional image to remove the background with non-uniform brightness distribution and obtaining the residual image. |
| Claim: |
3. The non-uniform image defect inspection method of claim 2, wherein in the step b2), the cut-off frequency is a frequency of the DCT coefficients on finishing a first order attenuation, wherein the DCT coefficients less than the cut-off frequency is remained, and the DCT coefficients higher than the cut-off frequency is eliminated through being set as 0. |
| Claim: |
4. A non-uniform image defect inspection method, comprising steps of: a) inputting an original two-dimensional image; b) eliminating a non-uniform background image from the original two-dimensional image by performing a DCT operation to obtain a residual image without the non-uniform background image; c) binarization segmenting the residual image to extract the defects from the residual image; and d) performing quantified value analysis on the binarization segmented defect regions and setting a quantifying threshold, wherein regions with a corresponding quantified value less than the quantifying threshold are regarded as acceptable regions and eliminated, and the residual defect regions are the inspection results. |
| Claim: |
5. The non-uniform image defect inspection method of claim 4, wherein the step b) further comprise the follows steps: b1) performing a 2-D DCT operation on the original two-dimensional image to obtain DCT coefficients of a frequency domain; b2) selecting a cut-off frequency from the frequency domain and eliminating high-frequency DCT coefficients higher than the cut-off frequency; b3) performing a 2-D IDCT operation to reconstruct a background image with a brightness distribution similar to the brightness distribution of the original two-dimensional image; and b4) subtracting the background image from the original two-dimensional image to remove the background having non-uniform brightness distribution and obtaining the residual image. |
| Claim: |
6. The non-uniform image defect inspection method of claim 5, wherein in the step b2), the cut-off frequency is a frequency of the DCT coefficients on finishing a first order attenuation, wherein the DCT coefficients less than the cut-off frequency is remained, and the DCT coefficients higher then the cut-off frequency is eliminated through being set as 0. |
| Claim: |
7. The non-uniform image defect inspection method of claim 4, wherein in the step d), the quantified value analysis is a quantified operation on SEMU values. |
| Current U.S. Class: |
382149/000 |
| Current International Class: |
06; 06 |
| Accession Number: |
edspap.20080107328 |
| Database: |
USPTO Patent Applications |
