Modification of the landweber method based on the conjugate gradient method to restore automotive radar images

The Landweber method is one of the simplest iterative image restoration methods. Although it may be a good choice for real time restoration of automotive radar images, this method suffers from the slow convergence and also the complexity of choosing its step sizes. In this paper, first a definition of the Landweber method on the basis of the line search methods is provided. Then, the convergence speed of the Landweber method with keeping its simplicity is improved. The stability of this method is also assured by controlling its step sizes. Applying this new method to simulated and real radar images shows its faster convergence and lower RMSE compared to the conventional Landweber method.