Real-Time Textureless 3D Object Tracking Based on Temporally Consistent Local Color Feature

Textureless 3D object tracking often fails because of cluttered backgrounds and motion blur, and slow tracking speed is difficult to achieve real-time tracking. To overcome these problems, this paper proposes a real-time textureless 3D object tracking algorithm based on temporally consistent local color feature. Firstly, color pixels are collected on the normal search lines that pass through the projected contour points of 3D model, and weighted mean of these pixels color is computed as local color features, which can enhance representation of color feature in cluttered backgrounds. The temporally consistent updating strategy is applied to the local color features, and local color features with similar foreground-background color are eliminated in the updating process to prevent tracking failure caused by similar foreground-background color. Then a novel energy function is defined based on the local color feature, and the differentials of the energy function with respect to the pose parameters are derived. Finally, the optimal pose parameters are searched using improved Gauss-Newton algorithm, in which a damping parameter is used to prevent convergence to a local minima, so that the accuracy of pose parameters and the tracking speed are improved. 7 real video sequences are tested to verify the performance of the proposed algorithm. Experiments demonstrate that the proposed algorithm can overcome the influence induce by cluttered backgrounds and motion blur, and implements real-time tracking without relying on parallel computing techniques. © 2020, Beijing China Science Journal Publishing Co. Ltd. All right reserved.