The performance of automatic target recognition in synthetic aperture radar images is greatly influenced by preprocessing, viz, despeckling and decluttering. In this work, a particle swarm optimization (PSO)-based adaptive wavelet packet transform is introduced for despeckling and decluttering of military targets including tanks, bulldozers, trucks, cars, cannons, and armored personnel carriers. The proposed method consists of two stages. The first stage removes speckle, and the second stage removes clutter with the aid of PSO to optimize the objective criteria, such as equivalent number of looks and signal to clutter ratio, respectively. The purpose of these methods is to enhance the target feature suitable for further processing. The proposed work has been tested on the moving and stationary target acquisition and recognition database and shows a remarkable performance over existing methods.