Algorithm 1: Motion Direction Estimation Using Modified Radon Transform with ρ range-based Summation

Convert blurred image into gray-level image Performing Hann windowing to remove the boundary artifacts Transform the image in step 2 from spatial domain to frequency domain using Fourier transform to obtain the image in frequency domain F(u,v) Compute the log of the power spectrum of F(u,v) in step 3 and remove the very high frequency components, as shown in Figure 7b Compute ρ range-based summation using Equation (10) Find the largest value (peak value), as shown in C of Equation (11), and indicate the direction of the dominant parallel line