The discriminative correlation filter (DCF) is one of the crucial visual tracking methods, and it has outstanding performance. Nevertheless, DCF-based methods have an unavoidable boundary effect, which results in poor tracking performance in an abrupt scene, such as fast motion or deformation. To address this problem, we propose a novel dynamic saliency discriminative correlation filter for visual tracking. In our approach, a response guided saliency map is constructed to introduce saliency information into the filter. The method effectively highlights the target by further increasing the number of positive samples to alleviate the boundary effect. We also investigate an effective multifeature integration method to extract the target feature by employing the Felzenszwalb Histograms of Oriented Gradients (fHOG) from each color space. Finally, we apply a novel update approach to prevent filter model degradation, which uses a temporal regularization term to update the filter model. Extensive experiments on the standard OTB-2015 benchmark validate that our approach achieves competitive performance compared to other state-of-the-art trackers. Moreover, we conducted an ablation study to evaluate the effectiveness of the components in our tracker.