Image Denoising is a method of removal or reduction of noises, that are incurred during the image capturing, image acquisition, and transmission due to the electronic and photometric sources. This paper is mainly focused to design an efficient architecture for removal of random-valued impulse noise from corrupted images. To achieve this, a good image denoising method property is that, it will remove noise while preserving edges is proposed. The proposed methodology is Decision-Tree-Based Denoising Method (DTBDM), which consists of decision tree based impulse detector, to detect the noisy pixel and to locate the edges by an edge preserving filter to reconstruct the intensity values of noisy pixels. Further, it is enhanced by using modified square root Carry Select Adder (SQRT CSA), in order to improve the execution time. Here, the serial adder is replaced with modified square root CSA to reduce computational time and area. So this technique can be used for many real time applications like medical imaging, scanning techniques, face recognition, etc.