Effects of rigid, affine, b-splines and demons registration on video content: A review

Image registration refers to transforming one image with reference to another. Image registration has become one of the key applications in medical imaging, as medical images captured by different devices often leads to distortion. In our present work, we have studied the effect of four different registration techniques, namely 1) rigid transformation based, 2) affine transformation based, 3) b-splines registration, 4) demons registration. Our proposed system defragmented a video content in multiple frames and applied a binning strategy prior to registration, which defined the middle frame of the bin being the target frame. Later, the registration effects were observed with the help of visual representation of registration frames, correlation value and mean structural similarity index measure (MSSIM) values between registered and unregistered frames. Our system also calculated the processing time for each registration technique in order to establish a comparative study among them.