Algorithm 2: The proposed spatiotemporal constraint algorithm.

Input: the blurry frame Bn ($n\ge$2), the motion-compensated frame In, the blur kernel k and the parameters λS and λT. Initialize the deblurred frame L = Bn.  While not converge do (1)Save the previous iterate: Lp = L. (2)With L fixed, solve the u-subproblem using (12) and (13). (3)With u fixed, solve the L-subproblem using (15).    If ${\Vert L-L_P\Vert }_2/{\Vert L_P\Vert }_2\le tol$ then    Break    End if  End while Output: the deblurred frame L.