Analysis of a printed complex image quality checking method of fabric cloth for development of an automated quality checking system
| Autor: | Saumendu Roy, Guoqing Hu, S M Jahangir Alam |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Accuracy and precision
Pixel business.industry Computer science Image quality media_common.quotation_subject ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Image registration 020206 networking & telecommunications 02 engineering and technology Image (mathematics) Software Signal Processing 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Quality (business) Computer vision Artificial intelligence Electrical and Electronic Engineering business Reliability (statistics) media_common |
| Zdroj: | Signal, Image and Video Processing. 15:195-203 |
| ISSN: | 1863-1711 1863-1703 |
| Popis: | In this research paper, the method of printed complex image quality checking for fabric cloth has been proposed for the development of an automated, printed image, quality checking system. As described in this paper, this quality checking system was implemented using graph matching and the pixel calculation algorithms, and the reliability was analyzed using HALCON software. The analysis of image and defect detection of image was performed by partitioning the images. First, the image registration area was automatically selected according to the regional characteristics of the image pattern; subsequently, the system performed fast image registration, based on the shape discovered through a graph matching method. Finally, the pixel calculation method was used in the specific image registration area to detect defects of the printed image in order to identify and ensure quality of the subsequently printed complex pattern on the fabric cloth. In this research, the proposed method was carried out on the image pattern of a selected registration area. The image registration for defective detection quickly, effectively, and efficiently detected the faults. The minimum and maximum errors were found as 0.16% and 8.4%, respectively, occurring within a 10% control. Whenever the error exceeded 10%, it was regarded as a quality defect of printed fabric cloth product; conversely, when the error fell below 10%, the product was deemed a qualified product. The higher measurement accuracy by the proposed method meets the needs of actual production by providing a new method of printing defect detection for the textile and garment industry printing sectors. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink