Digital images are used every day, and with the advent of mobile technology, the significance of image consumption and storage is growing day by day. Mobile devices have become the primary devices for capturing, storing, and consuming images replacing traditional cameras. In mobile devices, due to limited storage, it is most desirable to keep images compressed. Current image compression techniques treat two images independently and employ tools to remove only intra-image redundancy. However, this is not true for images captured and stored in mobile devices due to the user’s tendency to click multiple pictures with slight variation in composition, color, etc. For example, two images may differ only in foreground or background or maybe shifted slightly or may vary in zoom or rotation. We can obtain further compression by removing this interimage redundancy. In this thesis, I propose to research finding techniques to: • Discover redundancy between images by image matching and prediction techniques. • Employ database techniques to store image relations. • Efficiently compress images with inter-image redundancies removed. The main idea is to employ video compression techniques to predict an image using neighboring images and store only a prediction error. Prediction error can be further compressed by employing existing intra-image compression techniques. Research would also involve finding limitations in applying video compression techniques on images and modifying them for an image. The thesis also consists of developing an image gallery application that will support the encoding of images while adding, decoding images for viewing. Database techniques are used to store any change in the relation between images. The final goal is to achieve better compression in storing images when compared to traditional image compression techniques.