Scholarlink

<p>Manual evaluation of deformed potatoes remains a dominant method in agricultural quality control, despite being labor-intensive, time-consuming, and prone to subjectivity. To address this limitation, we propose an automated, non-destructive framework that combines classical image processing and statistical modeling to detect malformations and estimate usable weight in deformed potatoes. The system first extracts the potato contour using a combination of Canny edge detection and the Douglas–Peucker algorithm for shape simplification. Deformed regions are then identified using Graham’s scan algorithm based on geometric convexity analysis. A multiple linear regression model is employed to predict the usable weight by quantifying the defective area. Experimental results demonstrate that our method achieves a malformation detection accuracy of 96.0%, and outperforms deep learning-based segmentation (e.g., BiSeNetV2) by 1.75% in segmentation success rate. The predicted usable weight yields a root mean square error (RMSE) of 19.63&#xa0;g compared to manual measurements. This lightweight and interpretable approach offers a promising solution for real-time deployment in automated grading systems. Future work will focus on extending the method to multi-angle imaging and enhancing robustness under variable lighting and background conditions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Potato deformity detection and evaluation using Douglas-Peucker algorithm combined with graham algorithm

  • 期刊论文

  • 发表日期：

    2025年8月27日

查看全文
  • Fengnong Chen,
  • Kai Ren,
  • Ye Li,
  • Xiaofei Cheng,
  • Hongwei Sun

摘要

Manual evaluation of deformed potatoes remains a dominant method in agricultural quality control, despite being labor-intensive, time-consuming, and prone to subjectivity. To address this limitation, we propose an automated, non-destructive framework that combines classical image processing and statistical modeling to detect malformations and estimate usable weight in deformed potatoes. The system first extracts the potato contour using a combination of Canny edge detection and the Douglas–Peucker algorithm for shape simplification. Deformed regions are then identified using Graham’s scan algorithm based on geometric convexity analysis. A multiple linear regression model is employed to predict the usable weight by quantifying the defective area. Experimental results demonstrate that our method achieves a malformation detection accuracy of 96.0%, and outperforms deep learning-based segmentation (e.g., BiSeNetV2) by 1.75% in segmentation success rate. The predicted usable weight yields a root mean square error (RMSE) of 19.63 g compared to manual measurements. This lightweight and interpretable approach offers a promising solution for real-time deployment in automated grading systems. Future work will focus on extending the method to multi-angle imaging and enhancing robustness under variable lighting and background conditions.