A computer vision system for analyzing images of rough hardwood lumber

A computer vision system for locating and identifying defects on rough hardwood lumber in a species-independent manner is described. The system consists of three modules: a low-level module, a mid-level module, and a high-level module. The low-level module is a segmentation module that segments an input board image using a histogram-based thresholding method. The mid-level module eliminates small noise regions, merges similar adjacent regions, and computes region properties of merged regions. The high-level module identifies the defects present in each of the regions passed to it using a rule-based approach. The system is designed to be used in an automatic edger and trimmer for hardwood lumber in order to optimize the value of the hardwood boards processed.<<ETX>>