ABSTRAKSI: Kata Kunci : ABSTRACT: Currently the management of oil palm plantations are still having difficulty in getting information about the state of palm trees. The information obtained is still manually check directly. Besides palm oil plantations are also often used as collateral to secure loans by the owners and the lenders can not get definitive information about the condition of the oil palm plantation.
The design of the detection system of oil palm plantation-based method that uses digital image processing such as contrast enhancement, median filter, edge detection Canny, Bw open area, opening and closing. Labelling as feature extraction techniques contained in the image of palm oil plantations. Having obtained the characteristics of the image is then classed image using BackPropagation Neural Networks
System simulation is done with the help of tools (software) Matlab 2009. Samples are detected is the image of oil palm plantations in one district in Riau Province Kuantan Singingi district was taken using Google Earth and then processed with digital image processing to obtain information from the condition of the oil palm plantation. The output of this system in the form of information about the whereabouts and condition of oil palm plantations.
Tests on this system using five kinds of parameters, the variation of the median filter, variations in the value of Canny edge detection, variations in the value of Bw open area, the variation coefficient value of opening and closing the coefficient of variation values. Threshold value for the Canny edge detection is the best threshold value of 0.2. The best filter is the median value of 10, BW value of the best open area is 3, the coefficient is the second best opening, and closing coefficient is 4 and computing time is 16.41217298 second.Keyword: Detection of Palm Gardens, Digital Image Processing, Segmentation Shape, Back Propagation neural network, Information Condition