Abstract: The iterative form of the Super-Exponential （SE） algorithm has good convergence capability, but its complexity increases with the square of the equalizer length, which makes it unsuitable for hardware implementation. According to the properties of sparse underwater channels, a novel improved sparse Super-Exponential Iterative （SEI） blind equalization algorithm is proposed, which uses part of the SE algorithm to estimate the equalizer. After sparse processing, the equalizer is used to initialize the SEI algorithm. Less complexity and faster convergence rate is attainable with this method. The computer simulation results confirm the effectiveness of the proposed algorithm.