Image reconstruction is a difficult problem when the data is noisy, luckily regularization is a well known method for its solution. Variety of regularization techniques is used to obtain stable solutions. However they also involve challenges, a significant one of which is the selection of the regularization parameter. The parameter choice is very important in the sense that it balances the fidelity to data and prior knowledge in the solution. Our research is mostly concentrated on the selection of the regularization parameter when a non-quadratic regularizer is incorporated to the solution. Different applications such as ultrasound and radar imaging are in focus.