밤 박피율 분석을 위한 머신러닝 군집화 모델 개발 = A study on a machine learning clustering model for analysis of peeling rate of chestnut

국문 초록 (Abstract)

본 연구에서는 머신러닝 및 영상처리 기술을 이용하여 밤 박피율 분석 알고리즘을 개발하였다. 51가지의 2단 칼날박피 공정 조건별로 처리한 밤의 RGB 영상데이터를 획득한 후 이를 기반으로 머신러닝 군집화 모델을 적용하였다. K-means clustering 군집화 알고리즘을 적용하여 두 번의 군집화 및 랭킹 시스템으로 모델을 구성하였고 Jaccard similarity index를 이용하여 모델의 성능평가를 실시하였다. 첫 번째 군집화 단계에서는 원 RGB 영상에서 밤과 배경을 분리하는 과정을 수행하였고, 평균 Jaccard similarity index가 0.92로 나타났다. 두 번째 군집화 단계에서는 박피된 밤의 3가지 영역(외피, 내피, 과육)으로 분할하였고, 각 영역별 픽셀 개수와 영상의 총 픽셀 개수를 나누어 각 영역의 퍼센트 값을 도출하였다. 그 결과 외피, 내피, 과육의 Jaccard similarity index는 각각 0.96, 0.88, 0.74로 나타났다. 결과적으로, 본 연구를 통해 제안된 박피율을 정량적인 결과값을 분석할 수 있는 머신러닝 모델을 활용해 박피효율을 개선함으로써 국내 밤 산업의 생산 효율성 향상에 기여할 수 있을 것으로 생각된다.

다국어 초록 (Multilingual Abstract)

This paper proposes a machine learning-based clustering model to analyze the peeling rate of chestnuts. First of all, RGB image data of chestnuts treated using 51 two-knife peeling process conditions were acquired. The developed model was constructed with two stages of K-means clustering and ranking systems. The Jaccard similarity index for a performance evaluation of the models was calculated. The first clustering stage was a process of separating the chestnut and background from the original image. The average Jaccard similarity index of the first clustering step was 0.92. The second clustering step was the division process of the peeled chestnut into three regions (outer shell, inner shell, and kernel). The percentage of each region was derived by dividing the number of pixels in each region by the total number of pixels in the image. The Jaccard similarity indices of the outer shell, inner shell, and kernel were 0.96, 0.88, and 0.74, respectively. As a result, machine learning-based clustering with a ranking system was proposed to analyze whether the peeling rate affects the improvement of the peel efficiency and production efficiency in the domestic chestnut industry.

참고문헌 (Reference)

• 1 Korea Rural Economic Institute, "Vision of 2030/2050agriculture and rural sector in Korea"
• 2 K. P. Sinaga, "Unsupervised K-means clustering algorithm" 8 : 80716-80727, 2020
• 3 C. H. Son, "Trend and future of the domestic industry for chestnut" Korea Rural Economics Institute 77-79, 2003
• 4 H. Ding, "The application of artificial intelligence and big data in the food industry" 12 (12): 45111-, 2023
• 5 J. H. Kim, "Study on optimization of flame peeling system for chestnut(II)-Optimization of flame peeling process for chestnut-" 29 (29): 53-58, 2004
• 6 S. H. Park, "Research on internal quality of chestnut for non-destructive sensor development" 16 (16): 653-659, 2018
• 7 K. Srikaeo, "Peeling of gingers as evaluated by image analysis techniques : A study for pickled ginger process" 18 (18): 1387-, 2011
• 8 주석현 ; 김만조 ; 이욱, "Optimum condition of pellicle peeling using abrasion peeling machine and its characteristics among Chestnut varieties(Castanea spp)" 102 (102): 7-14, 2013
• 9 G. I. Ivchenko, "On the jaccard similarity test" 88 : 789-794, 1998
• 10 M. Gavahian, "Ohmic-assisted peeling of fruits : Understanding the mechanisms involved, effective parameters, and prospective applications in the food industry" 106 : 345-354, 2020
• 11 A. C. O. Ferraz, "Mathematical modeling of laser based potato cutting and peeling" 90 (90): 602-613, 2007
• 12 W. H. Su, "Imaging spectroscopy and machine learning for intelligent determination of potato and sweet potato quality" 10 (10): 2146-, 2021
• 13 F. Pallottino, "Image analysis techniques for automated hazelnut peeling determination" 3 : 155-159, 2010
• 14 Y. Kim, "Effects of acid and alkali treatment for chestnut peeling and its physical properties" 6 (6): 308-312, 1999
• 15 김현주 ; 정지희 ; 조유나 ; 진동은 ; 진수일 ; 김만조 ; 허호진, "Effect of daebo(Castanea crenata)inner skin extract on TMT-induced learning and memory injury" 45 (45): 661-665, 2013
• 16 J. H. Kim, "Development of chestnut peeling system" 22 : 289-294, 1997
• 17 황자영, "Correlation between physicochemical properties and peeling ratio of Korean chestnut" 33 (33): 190-197, 2017
• 18 K. Kug, "Applications of artificial intelligence technology in various industries"
• 19 K. Bandyopadhyay, "Application of artificial intelligence in food industry—A review" 5 (5): 145-152, 2021
• 20 A. Alwarid, "An absolute differences K-means clustering approach on determining intervals to optimize fuzzy time series Markov Chain model" 20 (20): 782-794, 2022
• 21 D. Kohli, "Advances in peeling techniques for fresh produce" 44 (44): e13826-, 2021