A new method for forming features of a textured image has been presented at the Moscow Technical University of Communications and Informatics (MTUCI), which can significantly improve the process of fabric sorting in recognition systems.
Fabric sorting is an important stage of quality control, which allows detecting defects and optimally using materials during cutting. Traditionally, this process was performed manually, which allowed detecting only about 70% of possible defects, such as pattern mismatch, loose threads, and color deviations. The new method, developed by MTUCI scientists, offers an automated approach to forming a mathematical model of textile material, which significantly increases the efficiency of sorting.
The key element of the new method is the creation of standards - samples with which fabric images are compared to identify defects.
Analyzing the quality of the fabric relative to the reference state allows highlighting significant features of the fabric structure, including analyzing the state of single weaves. In this problem statement, the fabric is considered as a two-dimensional periodic object.
To improve the quality of images in the sorting process, preprocessing methods were used, such as brightness correction, noise reduction, and image transformation for more detailed analysis.
The results obtained represent only one of the possible approaches to quality control of textile materials. In the future, MTUCI plans to develop systems capable of predicting and analyzing defects in detail at the production stage, which will open new horizons in the field of quality control automation.
Read more materials on the topic:
Electrically conductive textiles for "smart" clothing created in Russia
Russia starts producing fabric with "breathable" water-repellent membrane for the first time
Say "no" to washing: Novosibirsk scientists invented a self-cleaning fabric