Please use this identifier to cite or link to this item:
http://hdl.handle.net/10603/17861
Title: | Automatic defect detection using enhanced motif and non-motif algorithms in 2d patterned texture images |
Researcher: | Anitha S |
Guide(s): | Radha V |
Keywords: | Automatic Defect Detection Enhanced Motif Non-Motif Algorithms 2D Patterned Texture Images |
Upload Date: | 21-Apr-2014 |
University: | Avinashilingam Deemed University For Women |
Completed Date: | 01/10/2013 |
Abstract: | Defect detection of patterned fabric is a challenging problem in automatic inspection and is heavily used by the textile and garment manufacturing industries worldwide. It is an important task of quality control which helps to reduce rejection of fabrics and time spent on manual inspection. Over the past few years, many automatic computer aided defect detection systems that can detect defects from fabric images have been developed. Many of these systems are designed for non-patterned plain fabrics. However, the heavy use of patterns in the 21st century fashion world, defect detection in patterned fabrics has become more vital. This research work designs and develops defect detection systems for patterned fabrics.The proposed research methodology consists of three phases. The first step focuses on fabric image enhancement techniques. In this step, an input fabric image is enhanced by removing the impulse noise. For this purpose, an enhanced directional switching median filter is proposed.The second and third phase of the study focuses on the design and development of defect detection algorithms that identifies and locates defects in 2D patterned fabric images. The study considered both non-motif (Phase II) and motif (Phase III) based algorithms. In Phase II, the first technique proposed uses image data fusion technology which combines edge information obtained from Sobel edge operator, wave profiles and seam lines along with a dynamic threshold for defect detection. In the same phase, 12 wavelet-based techniques are also proposed. For this purpose, two wavelet variants, namely, optimal wavelet tree and Gabor wavelets are used. Optimal coefficients from these variants are selected using three algorithms. The first selection algorithm uses Vector Quantization and Principal Component Analysis, the second method uses Independent Component Analysis and the third performs a multiple projection-based selection combining the first two techniques. |
Pagination: | 222p. |
URI: | http://hdl.handle.net/10603/17861 |
Appears in Departments: | Department of Computer Science |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
01_title & initial pages.pdf | 572.98 kB | Adobe PDF | View/Open | |
02_chapter 1.pdf | Attached File | 762.21 kB | Adobe PDF | View/Open |
03_chapter 2.pdf | 211.17 kB | Adobe PDF | View/Open | |
04_chapter 3.pdf | 438.37 kB | Adobe PDF | View/Open | |
05_chapter 4.pdf | 382.54 kB | Adobe PDF | View/Open | |
06_chapter 5.pdf | 518.06 kB | Adobe PDF | View/Open | |
07_chapter 6.pdf | 259.79 kB | Adobe PDF | View/Open | |
08_chapter 7.pdf | 1.69 MB | Adobe PDF | View/Open | |
09_chapter 8.pdf | 20.54 kB | Adobe PDF | View/Open | |
10_chapter 9.pdf | 101.51 kB | Adobe PDF | View/Open |
Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
Altmetric Badge: