Automatic Twining Direction Recognition of Lockwire on Engine

Lockwires are often used as mechanical locking components for bolts on an engine. Inspecting its twining direction relative to the bolt is vital for ensuring an engine runs normally. However, the dominant manual inspection method is inefficient and limited by the professional skills of the inspector. In this paper, we propose a method that can be used to automatically determine the twining direction of a lockwire. First, a coarse-to-fine scheme for localizing the bolt is designed. Then, rough segmentation regions of the lockwire are acquired by a novel elongated shape descriptor. The entire area containing the lockwire can be obtained by connecting images from segmentation regions. An innovative skeleton tree and texture-based selection strategy are used to extract the lockwire centerline. An efficient vector fitting approach is proposed for computing the extension direction of the centerline. As a result, the twining direction can be identified via the extension direction and the bolt's position. The experimental results show that our method is very effective in determining the twining direction of a lockwire, with an accuracy reaching 95%. In addition, the results also show that our method can deal with low-quality images of lockwires at different angles.