In this paper, a method of data removal from images and then reconstruction of those images has been discussed. Images are reconstructed using B-Spline based interpolation of their Gray intensities and detailed qualitative and quantitative analysis has been carried out. The original and regenerated images are compared to generate graphical error profiles. Two techniques of B-Spline interpolation have been used. Error produced by both has been compared on the basis of parameters like mean error, standard deviation and variance. The same method is used for color data and the reconstruction results were impressive even at 2% of the original data. Apart from data-compression, this method can be used to regenerate images of good quality from low quality images and archived images which have deteriorated over time.