Name: Jaturong Som-ard
Title: Sugarcane Density Mapping and Estimation Yield Using UAV Image and OBIA Analysis
Estimating sugarcane yield is one of the direct ways to develop farmer’s income inThailand. This study was to develop a suitable classification model forsugarcane density mapping using UAV image and Object-Based Image Analysis(OBIA) technique and estimate yield. The dense maps were generated using MultiResolution Segmentation. OBIA, GE, and famer’s techniques were used to estimateyields, and these results were investigated using statistical methods. OBIA wassegmented using new rule-setting. The overall accuracy of OBIA that KK3 andUT12 were succeeded 91 and 85% which were most satisfactory. Estimating yieldwith OBIA was 197.63 (KK3) and 152.78 (UT12), and GE was 282.53 and 204.24.Farmer’s was 283.09 and 250.02, while actual yields were 183 and 148 tons. Theresults were investigated using the mean deviation of KK3 as 14.63, 99.53, and100.09, whereas UT12 as 4.78, 56.24, and 102.29, respectably. Thus, OBIA hadhigher accuracy to classify sugarcane density maps. OBIA gave better results ofestimating yield methods than another because the mean deviation values of OBIAwere more closed to mean value of actual yield. In addition, this technique ismost certainty to count a number of sugarcane stalks using dense maps. OBIA ishigh potential to segment structure of sugarcane canopy and leaf because it hasto consider spectral, spatial, and the textural image objects. Also, it is highefficiency to segment shadow, soil, and grass land. Therefore, this study helpsto develop varieties, and the farmers can use estimating methods for theirfurther improvement and decision making.