Cloud computing has been the latest topic of interest of experts from both the industry side and academic side. Cloud computing offers a variety of scalable services and cloud storage is a popular one among those. Apart from documents and other files, photo content is on the rise these days. Terabytes of photos are maintained by high-end cloud providers like Google, Microsoft and Amazon. Owing to this explosive growth of images, efficient compression mechanisms are required for storing images. JPEG is one of the popular schemes adopted for compression. Although it provides much compression, it is sufficient only for compressing as individual entities. On the other hand, this becomes an issue when it cannot handle redundancy between the images, which is required in photo album storage and retrieval. A few existing signals take into representative frames for accomplishing this task but severely fail if the photos are scaled differently or taken up indifferent illumination conditions. So the concept of compressing with local features was introduced and a few methods adopted a graph building approach. Efficiency wise, these techniques are good but fail on the context of computational time complexity. And hence the proposed algorithm based on Map-Reduce concept and SURF local features will try to overcome this time complexity involved, without reducing the performance. © 2015 IEEE.