追求環境保護（E，Environmental）、社會責任（S，Social）以及公司治理（G，Governance）的背景下[1]，實驗室開始專注於紡織產業的衣物布料回收研究。回收的材料包括常見的聚酯纖維（Polyester）、尼龍（Nylon）、棉（Cotton）以及聚酯纖維與棉的混紡等。過去，紡織產業通常需要依賴特定的化學技術進行材料鑑別，然後才能製造新的衣物。然而，現今實驗室已經能夠透過光譜變化的量測來確定衣物的材質[2]。本研究將採用Python語言開發，利用OpenCV影像辨識技術分析攝影機拍攝的衣物表面狀態，從中檢測出衣物的顏色，進一步協助化學分析。
本研究旨在利用在紡織產線上搭建的 Linux 系統架構，配備一組固定位置攝影機，拍攝紡織品流動的畫面，並透過影像處理技術分析衣物的顏色，將分析結果顯示於螢幕上。資料收集方面，我們與實驗室合作的紡織研究所以及舊衣回收所合作，從中獲取素色紡織品以及具有不同樣式的紡織品作為研究對象。實驗步驟包括以下幾個主要步驟：首先，從衣物中特定範圍抓取顏色像素，其中色彩空間主要使用HSV，將BGR像素轉換為色相（Hue）、飽和度（Saturation）、明度（Value），以定義和抓取顏色範圍；隨後，利用K-means主成分分群算法[12]，將特定範圍內的像素分群，並提取出其中的三個主要顏色；最後，將這三種主要顏色映射到三維空間中，計算它們之間的距離，以判定衣物的特性和顏色，例如素色（solid）、褪色或淡色（fade）以及具有不同樣式的衣物色（pattern）。透過這些步驟，我們旨在確定衣物的特性與顏色之間的關聯性。
分析結果顯示，成功開發了一個顏色辨識系統乃是透過Python程式和OpenCV影像處理技術，在色彩空間轉換和K-means主成分分群算法，能夠準確地抓取衣物的特定區域並識別出其顏色之外，可以用K-means的計算結果，知道三大顏色分群關聯性，並得知衣服的樣式類別。這一過程對於材料鑑定和化學分析提供了重要的基礎，使我們能夠更有效地處理和回收紡織品。

In the context of pursuing Environmental (E), Social (S), and Governance (G) responsibilities[1], the laboratory has begun to focus on research into the recycling of textiles in the clothing industry. The materials recycled include common fibers such as Polyester, Nylon, Cotton, and blends of Polyester with Cotton. Traditionally, the textile industry has relied on specific chemical techniques to identify materials before manufacturing new clothing. However, modern laboratories can now determine the composition of clothing[2] through spectroscopic measurements. This study will be developed using the Python language and OpenCV image recognition technology to analyze the surface condition of clothing captured by a camera, detecting the colors of the clothing to further assist in chemical analysis.
This research aims to utilize a Linux system architecture set up on the textile production line, equipped with a fixed-position camera to capture the movement of textiles, and analyze the colors of the clothing using image processing technology, with the analysis results displayed on the screen. In terms of data collection, we collaborate with textile research institutes and clothing recycling centers to obtain plain-colored textiles and textiles with different patterns as research subjects. The experimental steps include the following main procedures: first, extracting color pixels from a specific range of clothing, primarily using the HSV color space to define and capture the color range; then, using the K-means principal[12] component clustering algorithm, grouping pixels within the specified range and extracting the three main colors; finally, mapping these three main colors into three-dimensional space to calculate their distances to determine clothing characteristics and colors, such as solid, faded, or patterned.
Through these steps, we aim to determine the relationship between clothing characteristics and colors. The analysis results show that a color recognition system has been successfully developed through Python programming and OpenCV image processing technology. In addition to accurately capturing specific areas of clothing and identifying their colors, the K-means algorithm's results can reveal the correlation between the three major color clusters, and classify clothing style categories. This process provides an important foundation for material identification and chemical analysis, enabling more efficient handling and recycling of textiles.

Kulakova, I. (2018). The impact of Environmental, Social and Corporate Governance (ESG) practices on the financial performance of companies in emerging and frontier mar-kets (Master's thesis, TRITA-ITM-EX 2018:742, KTH Industrial Engineering and Man-agement, Industrial Management, SE-100 44 Stockholm).
Le, K. (2018). Textile Recycling Technologies, Colouring and Finishing Methods. Pre-pared for: Karen Storry, Senior Project Engineer, Solid Waste Services, Metro Vancouver. Vancouver, BC: UBC Sustainability Scholar.
Valladares, S., Toscano, M., Tufiño, R., Morillo, P., & Vallejo-Huanga, D. (2021). Per-formance evaluation of the Nvidia Jetson Nano through a real-time machine learning ap-plication. In Intelligent Human Systems Integration 2021: Proceedings of the 4th Interna-tional Conference on Intelligent Human Systems Integration (IHSI 2021): Integrating People and Intelligent Systems, February 22-24, 2021, Palermo, Italy (pp. 343-349). Springer International Publishing.
Intel, Willow Garage, & Itseez. (2000). OpenCV. Retrieved June 4, 2024, from https://github.com/opencv/opencv.
Mollon, J. D., Pokorny, J., & Knoblauch, K. (Eds.). (2003). Normal and Defective Colour Vision (ISBN-10: 9780198525301, ISBN-13: 978-0198525301). United States of Ameri-ca: Oxford University Press.
Pasquale, A. J. (2024). CC BY-NC-SA 4.0. Retrieved from https://cod.pressbooks.pub/physics1100/chapter/color/.
Valido, J. A., Cáceres, J. M., & Sousa, L. M. O. (2023). Physical and mechanical proper-ties of Ignimbrite from Arucas, Canary Islands. Environmental Earth Sciences, 82, 342. https://doi.org/10.1007/s12665-023-11024-9
Yan, F., Li, N., & Hirota, K. (2021). QHSL: A quantum hue, saturation, and lightness col-or model. Information Sciences, 577, 196–213. https://doi.org/10.1016/j.ins.2021.06.077
Chung, J., & Wark, S. (2016). Visualising uncertainty for decision support (DST-Group-TR-3325). Defence Science and Technology Group, Joint and Operations Analysis Divi-sion.
Refsnes Data. (1998). W3Schools. Retrieved June 15, 2024, from https://www.w3schools.com
SharkD. (2008, January 12). The RGB color model mapped to a cube. POV-Ray Object Collection. Retrieved from https://commons.wikimedia.org/w/index.php?curid=3375025
Shamrat, F. M. J. M., Tasnim, Z., Mahmud, I., Jahan, M. N., & Nobel, N. I. (2020). Ap-plication of K-Means Clustering Algorithm to Determine the Density of Demand of Dif-ferent Kinds of Jobs. International Journal of Scientific & Technology Research, 9(02), 2550. Retrieved from http://www.ijstr.org. (ISSN: 2277-8616)