Multimodal image registration finds widespread use in medical image analysis because of its usefulness in bringing together seemingly disparate sets of information from several imaging modalities. The major issue associated with image registration is that an algorithm works efficiently only for a given modality. In order to tackle this issue, we have proposed a novel multimodal image registration algorithm using Constrained Least Squares (CLS) that works for multiple medical modalities. The entire procedure of registration is based on the stipulation that the source and template images do not differ in zoom level but do differ in rotation and translation effects in the complex domain. To reduce the impact of noise and other inhomogeneities, a weight matrix is developed using prior knowledge of the local properties in the images. The least squares method is used to select the closed form solution to register the transformed image by estimating the displacement and rotation parameters. Many medical image registration algorithms have the limitation of quantitative analysis which we have addressed through Information Ratio (IR), Mutual Information Ratio (MIR), Lower bound Information Ratio (LIR), and Lower bound Mutual Information Ratio (LMIR).