The Comparison of the Edge Detection Methods in the Determination of Yarn Hairiness through Image Processing

The resolution, quality and speed of the cameras have improved enormously in recent years. The combination of camera advancements and the software industry offers significant opportunities. 

In this study, an image processing approach for the determination of yarn hairiness was presented. Yarn images taken under a microscope were examined in MATLAB software. 

Seven different edge detection algorithms were used in order to separate the hairs from the yarn body. Seven different textural properties of obtained yarn images were compared with Zweigle hairiness test results. The findings have indicated that yarn hairiness can be clearly detected from microscope images with a six-step algorithm. 

The first four phases are grayscale, double format, 2D median filtering and histogram-fitting, respectively. The fifth stage is the edge detection algorithm and the sixth stage is the use of textural parameters. When compared with the Zweigle hairiness results, the most obvious finding to emerge from this study is that the best appropriate technique for edge detection was the Sobel method, and the textural parameter to be used in the evaluation was the standard deviation of matrix elements.

*** Click here to download this article ***

Read More

The Relationship Between Subjective Pilling Evaluation Results and Detecting Pills and Textural Features in Knitted Fabrics

The digital image processing studies are used in order to eliminate problems of subjective pilling evaluation. However, these applications did not come to the desired point. The purpose of this research is to put forward with explanations about the reasons for the failures of previous studies in objective pilling evaluation. In this study, three issues were dwelled on. Firstly, data belong to original fabrics (0 turns) were taken into consideration. Secondly, data were standardized using min-max normalization with a feature scaling approach to compare different fabrics. For this process, data after pilling and results belong to original fabrics (0 turns) were taken together. Thirdly, knitted fabrics were separated into different categories according to formed pill types and characteristics after pilling processing. The results were evaluated in the most appropriate category according to the pill’s structure. Two sample fabrics containing appropriate structure and characteristics which were able to explain the three overlooked issues were used. In digital image processing made by paying attention to these mentioned three points, both pill parameters and textural features obtained from digital images were determined. The relationships between these parameters and subjective evaluation results were examined.

Source: Fibers and Polymers

Publisher: The Korean Fiber Society (KFS)

Original Language: English

Document type: Article / Restricted Access

Link: https://link.springer.com/article/10.1007/s12221-020-9552-1

Read More