Longan yield estimation using image processing technique and multiple regression model for decision support systems

• Main Authors: Veerachitt,V., Sangchyoswat,C.
• Format: Article
• Published: Chiang Mai University 2015
• Subjects: Multidisciplinary
• Online Access: http://www.scopus.com/inward/record.url?partnerID=HzOxMe3b&scp=84872253788&origin=inward; http://cmuir.cmu.ac.th/handle/6653943832/38055
• Institution: Chiang Mai University

Longan production system requires a quick and accurate estimation technique for planning. This paper presents the development of longan yield estimation method. This research identified types of land mapping units (LMUs) for longan production in Phrao district, Chiang Mai Province using spatial data, consisted of age of longan trees, irrigation systems and land gradient. The LMUs types suggested the determination of sampling areas for collecting data and taking photos of longan bushes. The visible wavelengths (RGB), and a range of relationship barrier patterns were determined for the conditions of image processing. The proportion of fruits to bushes was one factor among others gathered from farmer interviews. This information was applied to develop the longan yield model using a multiple regression equation. The equation could explain the relationship of various factors with the accuracy of 89% and the R2 of 0.79. The longan yield in each LMU was calculated by this equation also to generate an estimated yield map for describing the distribution of longan production at the crop scale. This information is crucial for the planning of longan production management prior to the harvesting. The planning can benefit to longan farmers for improving the efficiency of resources allocation and utilization.