Abstract: A novel algorithm named as adaptive correlation filtering tracking algorithm based on multi-resolution and multi-feature fusion is proposed to avoid the disadvantages of only using single Histogram of Oriented Gradients（HOG） feature, no scale change and no preprocessing of the videos with different resolutions in the Kernel Correlation Filter（KCF） algorithm. The proposed method fuses HOG, Color Name and Gray features to improve the characterization ability for the target. Although the fusion increases the complexity of computation, Principal Component Analysis（PCA） is applied to reduce dimensions to ensure the algorithm efficiency. The scores of the response for the scale filter and the scale pyramid are computed to obtain the optimal size of the target adaptively at the estimated position. Moreover, our proposed method utilizes the multi-resolution and multi-segment pre-process strategy to adjust the size of the tracking target of the videos with not only HR but also LR. The experimental results show that the precision and the success rate of the proposed algorithm on the OTB2015 dataset are 6.3% and 10.3% respectively higher than the KCF algorithm. The average speed of our algorithm reaches 38.16 FPS, which meets the real-time requirements. Compared with the remaining eight mainstream algorithms, the presented algorithm exhibits stronger tracking accuracy and robustness.