Merging Feature Method on RGB Image and Edge Detection Image for Wood Identification

The identification of timber types has become very important especially for identifying the name because this name of a timber, determining the quality and price of timber, determining the tax for particular timbers, determining the price of a timber based on the quality. Until now, these activities are still done manually in two ways: (1) using a magnifying glass to identify the cross-section of the timber, and (2) using a microscope, which is done in the laboratory basis. Some of the shortcomings of these manual methods include slowness, subjectivity, less accuracy and strongly dependent upon the human expert availability. This research attempts to resolve these problems by proposing a intelligent computational method with ANN. The proposed method utilizes merging feature on RGB image and edge detection image. Based on the experimental results, the proposed method gives the results of 100% and 95% accuracy of the training data and testing of data respectively. This research using 20 names of wood, who all come from Indonesia. And then the photo shoot of cross section is used for training and testing. For training using 2000 image, and use 500 images for testing. It is important to notice that proposed method only requires seven features of RGB images, and six features of edge detection image. Keywords— wood identification, ANN, RGB, edge detection, image processing, pattern recognition.