In this paper a generic method of Parameter Assistant Inpainting (PAI) is utilized in compression to exploit visual redundancy in color image. The entire image is analyzed and fit them with a model class based on features such as gradated, structural or textural significant in the image. This results in dividing the image region into featured blocks and nonfeatured blocks. The traditional compression techniques are employed for encoding of nonfeatured block while exemplar selection is performed on featured blocks and a set of parameters are derived. These parameters are then used for encoding the featured blocks. At the decoder end both nonfeatured blocks and parameters are decoded separately and restored with the help of PAI technique. The proposed methodology gives a high coding performance by saving bits for the featured region while a good perceptual quality is maintained. Thus the approach is unique in utilizing features in the image and employing PAI for reconstruction compared to current coding schemes such as JPEG 2000 and MPEG-4 AVC/H.264