Comparison of some features extraction method in wood identification

Identification of a timber usually use the general characteristics, such as smell, impression felt, weight, color, and others. Key features include color, weight, felt the impression, and others. This method has low accuracy, and also requires a fairly long experience. Another way is to record the microscopic characteristics. It also requires quite a long time to recognize the type of wood. This type of wood is very influential on the price and usefulness. The identification of wood species (name of timber) using microscopic image-based neural network have ever done researcher before. But this time researcher want to compare the four feature extraction methods, that is 1) black and white image without the threshold value, 2) black and white image with the threshold value, 3) edge detection, 4) RGB. The best results are given of each method is 28%, 36%, 40%, and 88%. Of the four existing methods are still not capable of exceeding 98%.