Title: Customizing Progressive JPEG for Efficient Image Storage
Advisors: Luis Ceze and Xi Wang
Abstract: Modern image storage services, especially those associated with social media services, host massive collections of images. These images are often replicated at many different resolutions to support different devices and contexts, incurring substantial capacity overheads. We describe a prototype system design, ShoeBox, to investigate the effectiveness of replacing redundant multi-resolution image storage with progressively encoded JPEG images. After evaluating ShoeBox’s design and performance, we switch to an approach better suited to the capabilities of progressive JPEG which instead resizes images at request time. We describe methods of handling the inefficiencies associated with resizing images on the fly, and propose repurposing the progressive JPEG standard and customizing the organization of image data to reduce the bandwidth overheads of dynamic resizing. We show that at a PSNR of 32 dB, dynamic resizing with progressive JPEG provides 3× read data savings over baseline JPEG, and that progressive JPEG with customized encode parameters can further improve these savings to almost 6×. Additionally, we characterize the decode overheads of progressive JPEG to assess the feasibility of directly decoding progressive JPEG images on energy-limited devices. Our approach does not require modifications to current JPEG software stacks.