Image Approximation using Singular Value Decomposition vs the Spatial Information

In this paper it has been shown that Spatial Information can be a measuring tool for complexity of an Image. If the complexity of an image in terms of visual perception is high then the Spatial Information is also high and vice-versa. An image can be fairly approximated applying the Singular Value Decomposition technique with requisite number of singular values and after analysis, an equation is also proposed to quantify the number of singular value required to attain certain relative error threshold w.r.t. spatial information of that image. It has also been shown that spatial information of an image is sensitive to singular values of intermediate order rather that of higher order. So, more the spatial information an image has, more of the singular values are needed to approximate that image.