CLASSIFICATION OF RICE PLANT PESTS USING HAAR-LIKE FEATURE AND ADABOOST ALGORITHM

Efforts to monitor pest populations at a rice plant site are important because based on information on the type and number of pests attacking rice crops, a suggestion of controlling can be developed early so that potential losses resulting from pests can be suppressed. Therefore a process is needed to identify and classify the pests that attack and harm the rice plants. In this research will be designed rice pest classification using image processing where in its processing using image from stem borer (moth). Feature extraction of positive samples (pest image of moths) and negative samples (non-pest image) using Haar Like Feature. While in the process of classification into a class of moths and not moths using Adaboost algorithm by applying cascade classifier to get a strong characteristic. The observed variable is the error rate generated in the process of pest classification of moth and non pest. From the test result on positive samples obtained identification rate of true positive (TP) = 90%, while false positive (FP) = 20%. For negative sample test (non pest image) obtained true negative (TN) = 80%, while false negative (FN) = 20%. From the test result of positive sample and negative samples obtained the accuracy of pest moth identification results of 85%