A Comparison of Corner and Saliency Detection Methods for Power Line Detection

Issuing an alarm for power lines continues to be an important issue for low altitude aircraft safety. In this study, edge and gradient based methods are applied and compared according to their detection performances (in terms of Accuracy, F-Score, Speed and False Negative). It was observed that; despite being used genuinely for corners, HARRIS method provides significantly superior accuracy and F-Score for infrared images, and SIFT method provides superior accuracy and F-Score for visible light images. Interestingly, FAST method provides superior speed and false negative rate for both infrared and visible light images.