High Visual Quality Particle Based Frame Rate Up Conversion With Acceleration Assisted Motion Trajectory Calibration

Traditionally, frame rate up conversion can be realized by either bidirectional or unidirectional approaches. Bidirectional approach suffers from motion ambiguity. However, unlike unidirectional approach, bidirectional approach does not suffer from holes and occluded regions. In this article, the particle-based frame rate up conversion algorithm is proposed by combining hierarchical motion vector field construction stage and particle-based motion trajectory calibration stage. First of all, color-oriented adaptive cost function is applied in the hierarchical motion vector field construction stage to retrieve initial motion vectors. Second, the retrieved motion vectors are memorized and calibrated by minimizing the distance between interpolated block and particle trajectory intersecting points. Besides the minimization, joint cost-function evaluation further prevents untrue motion vectors from being selected. According to the experiment results, proposed pixel-based frame rate up conversion outperforms existing works by 0.9-3.9 dB. On the other hand, the computation complexity of proposed work is less than existing bilateral approaches by 5-100, while it is larger than the unidirectional approach by 3-100, according to the target video resolution. This concludes a high visual quality frame rate up conversion design.