Positron Emission Tomography (PET) images reconstructed without adequate scatter corrections introduce noise and degrade image contrast. In commercial imaging systems, misalignment between computed tomography (CT) and PET images can introduce biases in the activity distribution. Recently, several reconstruction algorithms have been proposed, which made direct use of single scattered photons in the activity reconstruction. However, the realistic dataset contains single and higher order scattered photons, and current scatter reconstruction methods do not distinguish them. In this study, a novel reconstruction algorithm that is capable of processing higher order scattered photons was developed. A restricted attenuation correction method was created to avoid overcorrecting for scattered photons. The simulation outcomes have shown that the proposed methods can, under ideal energy resolution, reconstruct images that are qualitatively and quantitatively better than those obtained using existing algorithms, and that the methods show promise for use under more realistic clinical conditions.