KEMU

Choosing a food becomes something important for sufferers of certain diseases. However, choosing a food is a problem for people who taste a food for first time or tourists who are visiting a country for first time. To overcome these problems, research needs to be done to identify / classify a food image. The Haralick and CIE Lab Color Moments features are proven to produce good features for classification cases. The Learning Vector Quantization method is also an alternative for classification process. Based on the k-fold cross validation with k = 10 and accuracy as evaluation method, the maximum accuracy is 0.642051 with learning rate parameter value is 0.2, the learning rate multiplier is 0.8, the m value is 0.1, the epsilon value is 0.4, maximum iteration is 10 and minimum learning rate is 0.000001. This result shows that food image classification based on Haralick feature extraction and CIE Lab Color Moment using Learning Vector Quantization produces fairly good accuracy. In addition, the use of both texture (Haralick) and color features (CIE Lab Color Moments) has an effect on the results of accuracy. This is indicated by all the test results which show that the highest accuracy results are achieved using texture and color features.