บทความวิจัยการประชุมวิชาการระดับชาติมหาวิทยาลัยนอร์ทกรุงเทพพ.ศ. 2568

การตรวจหาความสุกของมังคุดระดับที่ 4 โดยใช้การประมวลผลสีเปลือกมังคุดตาม ระบบสี HSV

The 4th level mangosteen ripeness detection using mangosteen rind color processing based on HSV color system

ผู้จัดทำ: สรินทร์รัชต์ เกตุขาว, เตชค์รูสิณป์ เพียซ้าย
ระดับการศึกษา: -
สาขาวิชา/กลุ่มสาขา: ด้านสหวิทยาการ วิทยาศาสตร์และเทคโนโลยี
0 เข้าชม 0 ดาวน์โหลด
ดาวน์โหลดเอกสาร

บทคัดย่อ (ภาษาไทย)

การตรวจหาระยะความสุกของมังคุดที่เหมาะสมถือเป็นสิ่งจําเป็นสําหรับการเตรียมการขนส่งมังคุดปริมาณมากออกสู่ตลาดภายในประเทศเพื่อให้การขนส่งได้ส่งถึงพอดีกับการบริโภค จากระยะเวลาการเก็บเกี่ยวจนถึงการ บริโภคมังคุด พบว่า การคัดแยกความสุกของมังคุดระดับที่ 4 เหมาะสมสําหรับการเตรียมขนส่ง ดังนั้น บทความนี้ เสนอการตรวจหาความสุกของมังคุดระดับที่ 4 โดยใช้การประมวลผลสีเปลือกมังคุดตามระบบสี HSV โดยภาพมังคุด ที่ถูกระบุโดยผู้เชี่ยวชาญจะถูกแปลงระบบสีจาก RGB เป็น HSV และทําการกําหนดแต่ละช่วงสีของ HSV จากสี เปลือกมังคุดที่มีช่วงระหว่างค่าสูงสุดถึงค่าต่ําสุดในแต่ละสี HSV โดยเป็นช่วงสีที่ทําให้ครอบคลุมพื้นที่สีเปลือกมังคุด มากที่สุดเพื่อใช้เป็นค่าอ้างอิงปกติในการตรวจหาแบบอัตโนมัติ การพัฒนาการตรวจหา ใช้เครื่องมือของโปรแกรม Visual Studio Code โปรแกรมไลบรารีคําสั่ง OpenCV และโปรแกรมภาษาไพธอน เวอร์ชัน 3.10.9 ผลการทดลอง กับภาพมังคุดที่มีความหลากหลายทั้งขนาด ปริมาณ และระยะความสุก จํานวน 50 ภาพ พบว่า สามารถตรวจหา ความสุกของมังคุดระดับที่ 4 ได้ความแม่นยําร้อยละ 91.29

Abstract (English)

Detecting the appropriate stage of mangosteen ripeness was essential for preparing the transportation of many mangosteens to the domestic market and consumer suitably. A time of harvest until the consumption of mangosteen found that the mangosteen ripeness selection at the fourth level was suitable for preparing the transportation. Therefore, this article proposed the fourth level mangosteen ripeness detection using the mangosteen rind color processing based on the HSV color system. The mangosteen images that the experts identified transformed from RGB to HSV color system and each HSV color determined the ranges of mangosteen rind color with the maximum and minimum values. These color ranges covered the largest area of mangosteen rind color to be used as a default value for automatic detection. The development of the detection used tools of the Visual Studio Code, OpenCV library, and Python version 3.10.9 programs. The experimental results with 50 mangosteen images with various sizes, quantities, and ripeness stages showed the detection accuracy of the fourth level mangosteen ripeness at 91.29%.