Adaptive Target Detection Algorithm Based on Correlation Filtering

Abstract

Correlation filtering is a fast and robust signal detection and processing algorithm. However, in the field of images processing, scale and rotation variation are important issues for correlation filtering algorithms. This paper proposes a detection method based on correlation filtering, which uses axis of symmetry and circumscribed rectangles to estimate the rotation and scale of the target. Firstly, using the HSV model to separate the colors to be found in the original image. Then, the axisymmetric shell intersection method is proposed to solve the symmetry axis and the circumscribed rectangle of the object. Finally, a correlation filter is used to solve all the areas that are likely to be objects. The response value, the position with the highest response value is the object position. The algorithm performs experiments on a set of artificially calibrated image sequences. Experiments show that this method can achieve better detection results when the number of training samples is small.