Generation of future image frames using optical flow

This research work presents a novel approach for generation of future image frames using optical flow method that estimates velocity of all the pixels in both axes of images in an image sequence. Direction and magnitude of velocity vector field helps in finding the changes in term of pixel intensity from one image to another. The pixel intensity change from one image to another is modeled Takagi-Sugeno fuzzy model (TSFM) in both directions. This network predicts velocities of each pixel and then corresponding pixels intensities are mapped to their new position. The resulting scheme has been applied successfully on an image sequence of landing fighter plane. The proposed approach is able to generate upto ten future image frames successfully. For the quality assessment of future generated images Canny edge detection based Image Comparison Metric (CIM) and Mean Structural Similarity Index Measure (MSSIM) is used. All the ten future generated images have been compare qualitatively against the test images and the results found are encouraging.