Quantum fast corner detection algorithm

Abstract

Corners are very important local features for an image that can be widely used in many image processing tasks such as object detection, image recognition and data compression. However, current corner detection algorithms are complex and time consuming. In this paper, a new quantum fast corner detection algorithm is proposed taking full advantage of quantum parallelism. The algorithm is implemented in three steps: the ‘nucleus’ is defined and its neighborhood pixels are determined in the first step. Two thresholds are selected in the second step, and corners are obtained in the third step. The first and second steps are the same as the classical corner detection algorithm. The third step is divided into two stages. In the first stage, the differences between grayscale values of the nucleus and pixels in the neighborhood are calculated, followed by the comparison of those differences with the ‘intensity threshold,’ then is the quantum measurement on the comparison results, and finally the measured results are organized into an array. In the second stage, the array is used for counting, followed by the comparison of those counted results with the ‘accuracy threshold,’ and finally the corners are obtained. It is worth noting that through the quantum–classical–quantum mode, quantum resources can be reduced significantly. The analysis in the proposed quantum fast corner detection algorithm shows that the time complexity and quantum delay of the algorithm do not increase with the increase in the size and number of images, and its time complexity is exponentially lower than that of the classical fast corner detection algorithm.