Pores Segmentation Based on Active Contour Model for Automatic Wood Species Identification

Wood species identification based on computer vision is an alternative solution in early-stage identification to obtain faster accuracy results. Hence, the development of computer-based identification is highly required. Each wood species has its unique anatomical characteristics. One of these characteristics is the pore. In order to obtain the pores object from the surrounding environment in a wood image, a segmentation is required. The aim of this research is to propose an automatic algorithm based on Active Contour model for wood pores segmentation in the wood species identification. In this research, three Active Contour segmentation models were used to obtain pores objects, i.e., Chan Vese (CV), Region Scalable Fitting (RSF) and Geodesic Active Contour (GAC) model. Nine commercial wood species will be used as a dataset to be segmented with different pore sizes. The result showed that the Chan Vese (CV) model is the best model for the pores segmentation than the RSF and GAC model with the highest IoU average value of 0.8883, the lowest iteration (N = 80), and the fastest segmentation running time (T) which ranges from 54.01 to 62.64 seconds.