Research On Correlation Filter Tracking Algorithm For Feature Preferred And Coordinate Correction

Object tracking has been a very important research content in computer vision,and related researchers have done a lot of work.However,there are still many problems in the field of target tracking.In order to achieve a more robust tracking process,this article focuses on the traditional kernel correlation filter tracking framework,by combining existing tracking algorithms and deep convolutional network knowledge.The main work is as follow.Aiming at the limitations of single features and insufficiencies update of target templates and feature appearance templates in traditional kernel tracker,a multi-channel feature and preferred parallel update for kernel correlation filter tracking is proposed.The multi-channel feature extraction method is adopted which means the upper branch uses the convolutional neural network to extract the depth features,and the lower branch combines the features of HOG and CN for training and tracking.A novel preferred parallel update method is proposed for updating the target templates and feature appearance templates.In the current frame,the maximum response value in both branches is considered as the optimal target position.In the next frame,the templates of the two branches are updated simultaneously with the parameters of the optimal position of the previous frame until the end of the tracking.The optimal parallel update of multiple branches makes up for the deficiency of single branch update.Experiments show that this algorithm can achieve more robust tracking result under different challenge factors.Aiming at the limitations of single-layer depth feature and the deviation of the final target prediction value,a multi-layer deep fusion features and coordinate correction for kernel correlation filter tracking is proposed.The multi-layer deep fusion features is adopted which means the output features of the three layers of Conv3-4,Conv4-4,and Conv5-4 of VGG19 are extracted and fused through pooling and splicing operations for training and tracking.Coordinate correction strategy is proposed.The correction factor and four transformation factors are obtained from the ground true value of the first frame target and the predicted position of the first target,and the optimal target position is obtained by zooming and panning operations.In the next frame,based on the predicted target position,four transformation factors and the optimal target position are determined.The subsequent frame template is updated with the parameters of the optimal position until the end of the tracking.Experiments verify that the algorithm can cope with different challenge factors to achieve a more robust tracking process.