Research On Frame Rate Up-conversion Technology Based On Motion Vector Field Smoothness Constraint

With the development of high-end display devices, we need to improve the existing frame rateof video program to obtain a better visual effect. Besides, in bad network condition, we can reducethe frame at the sending end so as to reduce the data transmission traffic. Then we upgrade theframe rate of the receiving video at the receiving end.Frame rate up-conversion is mainly used toimprove the video frame rate, so it has broad application prospect. After the elaborate of basicconcept and principle of the video frame rate up-conversion technology, this paper mainly researchthe frame rate up-conversion algorithm based on motion vector field smoothness constraint from thefollowing aspects:(1) After a detailed introduction of3-D Recursive Search (3DRS) algorithm, a new improvedtechnology is presented in the light of its existing problems. Compared with the original3DRSalgorithm, the improvements can be seen from the following three aspects: First, adjust the selectionof motion vector candidates, so as to eliminate the time candidate vectors which may introduceerror vectors, and increase the space candidate vectors; Second, add a reverse motion estimation toaccelerate the converge of motion vector. Third, a new matching criterion is used to improve theprecision of motion estimation in the object edges. Experimental results show that the algorithm inthis paper effectively improves the accuracy of motion estimation and the results achieve bettereffects both subjectively and objectively.(2) A new frame rate up-conversion technology is put forward. Optimize the3DRS algorithm byadd full search in the reverse motion estimation to improve the motion estimation accuracy of smallobjects with independent movement. Calculate the motion vector passed every pixel in interpolatedframe by the moiton vector field of current frame, and the block of interpolated frame as thebenchmark, the average of all moion vectors through each bolck as the initial motion vector, thenusing small scope weighted bi-directional motion estimation for correction. For inaccurate motionvector, by making full use of the8contiguous blocks of the inaccurate motion vector to smooth theexception block. In the stage of motion compensation, this paper uses an adaptive compensationmethod for motion compensation. The experimental results show that the algorithm has good effect,strong robustness.The experimental results show that the algorithm has good effect, strongrobustness.