Circular Hough Transform and Local Circularity Measure for Weight Estimation of a Graph-Cut Based Wood Stack Measurement

Abstract

One of the time consuming tasks in the timber industry<br/>is the manually measurement of features of wood stacks.<br/>Such features include, but are not limited to, the number<br/>of the logs in a stack, their diameters distribution, and their<br/>volumes. Computer vision techniques have recently been<br/>used for solving this real-world industrial application. Such<br/>techniques are facing many challenges as the task is usually<br/>performed in outdoor, uncontrolled, environments. Furthermore,<br/>the logs can vary in texture and they can be occluded<br/>by different obstacles. These all make the segmentation of<br/>the wood logs a difficult task. Graph-cut has shown to be<br/>good enough for such a segmentation. However, it is hard<br/>to find proper graph weights. This is exactly the contribution<br/>of this paper to propose a method for setting the<br/>weights of the graph. To do so, we use Circular Hough<br/>Transform (CHT) for obtaining information about the foreand<br/>background regions of a stack image, and then use this<br/>together with a Local Circularity Measure (LCM) to modify<br/>the weights of the graph to segment the wood logs from the<br/>rest of the image. We further improve the segmentation by<br/>separating overlapping logs. These segmented wood logs<br/>are finally scaled and used to acquire the necessary wood<br/>stack measurements in real-world scale (in cm). The proposed<br/>system, which works automatically, has been tested<br/>on two different datasets, containing real outdoor images<br/>of logs which vary in shapes and sizes. The experimental<br/>results show that the proposed approach not only achieves<br/>the same results as the state-of-the-art systems, it produces<br/>more stable results.