| URN | etd-0720126-161331 | Statistics | This thesis had been viewed 4 times. Download 2 times. |
| Author | Kuei-Chen Lin | ||
| Author's Email Address | niwahayon@gmail.com | ||
| Department | Institute Of Mechanical Engineering | ||
| Year | 2025 | Semester | 2 |
| Degree | Master | Type of Document | Master's Thesis |
| Language | zh-TW.Big5 Chinese | Page Count | 63 |
| Title | A Study on Machine Vision-Based Measurement and Quality Grading of Log-Cultivated Shiitake Mushrooms | ||
| Keyword | |||
| Abstract | Log-cultivated shiitake mushrooms (Lentinula edodes) are a high-value agricultural product in Taiwan. However, their market grading has long relied on manual visual inspection and subjective experience, which often leads to inconsistent standards, difficulty in quantification, and limited grading efficiency. To improve the objectivity and consistency of quality evaluation, this study develops a machine vision-based dual-side measurement and grading system for shiitake mushrooms. The proposed system integrates geometric features of mushroom caps from the front side with cap margin thickness information from the back side to establish a quantitative grading framework applicable to practical use. In terms of methodology, the front-side images are first converted into grayscale and processed using Otsuˇ¦s automatic thresholding combined with morphological operations to extract the region of interest. Least-squares ellipse fitting is then applied to obtain geometric parameters, including the major axis, minor axis, equivalent diameter, and ellipse area. These measurements are further calibrated from pixel units into real-world physical dimensions (millimeters). For the back-side images, RGB images are converted into HSV color space to enhance contrast among different tissue structures. An adaptive region of interest (ROI) combined with K-means clustering is employed to segment the cap margin region. Subsequently, polar coordinate transformation and radial pixel integration are introduced to effectively measure the thickness and completeness of irregular annular structures, thereby improving measurement stability and accuracy. A total of 218 samples were analyzed in this study. The results show that the average major axis, minor axis, and cap margin thickness are 68.86 mm, 61.24 mm, and 6.79 mm, respectively. Statistical analysis reveals a strong positive correlation between cap size and cap margin thickness, with a correlation coefficient of approximately 0.86ˇV0.87, indicating good consistency and representativeness of the dual-side measurement features. In addition, most samples are concentrated in medium and large size categories, suggesting that size serves as a primary grading criterion, while thickness provides further differentiation of quality levels. Based on these findings, a three-dimensional grading framework of ˇ§size ˇŃ thickness ˇŃ shapeˇ¨ is established. By incorporating size classification (S/M/L), thickness grading (A/B/C), and shape correction (G/N/P), the proposed method effectively enhances grading precision and decision accuracy, and demonstrates strong potential for practical application. In conclusion, this study successfully transforms traditional subjective evaluation of shiitake mushroom quality into a measurable, repeatable, and engineering-oriented data model. The proposed approach not only improves grading consistency and efficiency but also provides a solid technical foundation for automated grading and the advancement of smart agriculture in the shiitake mushroom industry. |
||
| Advisor Committee | |||
| Files | indicate access worldwide | ||
| Date of Defense | 2026-07-14 | Date of Submission | 2026-07-20 |