This paper presents a novel method for Visual Object Tracking (VOT), which treats some challenging problem such as significant changes in appearance caused by occlusion and variations in illumination. The proposed approach combines the Deep Convolutional Neural Networks (DCNN), Discrete Cosine Transform (DCT), Histograms of Oriented Gradients (HOG) features and HSV energy condition. Firstly, an HSV-based energy condition is employed to enhance the learning process by incorporating both RGB and HSV color bases. Instead of using the image template, the technique utilizes the coefficients of the image DCT to handle high saturation images in the Convolutional Neural Networks (CNN’s) input. The Inverse Discrete Cosine Transform (IDCT) is used to extract the CNN features. Secondly, the multichannel correlation maps generated by the CNNs are utilized to determine the target position. This is achieved by combining convolutional features. Newton's method is also employed in this process to enhance the long-term memory of the target's appearance and assist in recovering from tracking failures. The updating parameter for the correlation filters is calculated as the highest value among the output maps generated by correlation filters using convolutional features derived from the HOG features of the image template. Finally, the results obtained undeniably prove that the proposed method surpasses most recent tracking techniques

Convolutional neural network, Discrete Cosine Transform (DCT), Correlation filter, Visual tracking, Newton’s method