Noise is one of the main factors that impact the performance of optical flow where the result in motion vector of optical flow is degraded. Many areas in advance require a result of optical flow as a preprocessing such as super resolution image reconstruction, robot vision, motion estimation, edge detection, motion tracking and etc. Then, the accuracy in the result of motion vector from optical flow is very importance. This paper introduces the robust model of optical flow by using bidirectional confidential technique together with bilateral filter for global based spatial correlation where the efficiency of the proposed model is evaluated by Peak Signal to Noise Ratio (PSNR) under noisy condition.