Knit Fabric Dyeing Evaluation Using Aspect-Ratio Image Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for evaluating dyeing quality of knit fabrics through image analysis fail to accurately match human visual evaluations due to the extraction of non-uniform dyeing caused by factors other than oriented molecular crystal unevenness, leading to inconsistent and incorrect results.
Innovation Solution
An image processor is employed to extract high-contrast areas and then further refine these areas based on aspect ratio and size in the wale direction to identify uneven dyeing, which is long and narrow, and connect adjacent areas within specific distance criteria, while correcting for image orientation and removing noise and mesh patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-contrast areas are extracted to identify uneven dyeing, then detection sensitivity is improved, but false detection increases due to inclusion of non-dyeing unevenness
Solution Approach 1:
The patent applies local quality by imposing different extraction conditions on different types of areas. Specifically, uneven dyeing areas are extracted with aspect ratio ≥ 2 and area ≥ 50 pixels, while non-uniform dyeing areas are extracted with aspect ratio < 2 or area < 50 pixels. This differentiated local extraction criterion enables the system to distinguish between actual uneven dyeing and other surface defects, resolving the contradiction between detection sensitivity and evaluation accuracy.
2Productivity
If simple high-contrast area extraction is performed, then processing speed is improved, but evaluation consistency deteriorates due to inclusion of irrelevant features
Solution Approach 1:
The patent applies parameter changes by introducing specific quantitative thresholds for aspect ratio (≥2) and area (≥50 pixels) to differentiate uneven dyeing from other defects. These parameter changes transform the evaluation process from subjective visual assessment to objective automated measurement, ensuring consistent evaluation results while maintaining processing efficiency through algorithmic extraction.
3Extent of automation
If automated image analysis is implemented, then evaluation objectivity is improved, but matching with human visual evaluation deteriorates due to extraction of irrelevant features
Solution Approach 1:
The patent applies local quality by extracting different types of high-contrast areas with different criteria. Uneven dyeing areas (aspect ratio ≥ 2, area ≥ 50 pixels) are extracted to match human visual evaluation, while non-uniform dyeing areas (aspect ratio < 2 or area < 50 pixels) are extracted separately and excluded from the evaluation score. This selective local extraction enables automated analysis to accurately replicate human evaluator perspectives.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An object of the present invention is to allow an evaluation result obtained by the image analysis to match to an evaluation result obtained by the visual evaluation of an evaluator when the dyeing quality of the knit fabric is evaluated by analyzing the image obtained by capturing an image of a dyed knit fabric. An image processor 7 is configured to extract an uneven area corresponding to uneven dyeing from an image of a dyed knit fabric S. The image processor 7 performs: an edge detecting process of detecting an edge included in the image; a first extracting process of extracting a high-contrast area in which a gradient of the edge is equal to or higher than a predetermined value; and a second extracting process of extracting an area in which an aspect ratio is equal to or higher than a predetermined value as the uneven area from among the high-contrast area. The aspect ratio is a ratio of a size of the area in a course direction with respect to a size of the area in a wale direction.