Personal photo album compression and management

Image Insertion Using Depth- and Topology- Constrained Minimum Spanning Tree for Compressed Image Sets

“…Wu et al [21] proposed that for a to-be-deleted image if it is a leaf vertex it will be directly removed, or its child vertices will be re-encoded by using its parent vertex as their new parent vertex. In [23], each to-be-deleted image was only marked as "deleted" but without removing its information at all in actual from its compressed image set, unless it was a leaf vertex or all of its descendants had been marked as "deleted". A parent vertex determination method [36] was proposed to obtain a new reference image for each to-be-remained image, and then encoded these to-be-remained images using their respective new reference images to generate the final compressed image set after deletion.…”

“…To address the issue of image insertion, to the best of our knowledge, only in [21] and [23] this problem is partially alleviated for compressed image sets. In order to add a new picture to a compressed image set, it is necessary to find a compressed image from the compressed image set, which is then acted as the reference in encoding this new shot.…”

“…In order to add a new picture to a compressed image set, it is necessary to find a compressed image from the compressed image set, which is then acted as the reference in encoding this new shot. For the way of identifying that reference image, the methods in [21] and [23] were rather different, where in [21] the root of the existing compressed image set was directly used to be the prediction reference for all the new photos, whereas in [23] an exhaustive search was executed to seek out a compressed image with the minimum prediction cost in encoding every new photograph.…”

“…Currently, how to realize image insertion has been especially focused on in [21] and [23]. In [21], Wu et al introduced that each new picture was immediately inserted by encoding it only as a child vertex of the root of an existing compressed image set.…”

“…There is no need for this method to consider the other images of the compressed image set. In [23], by searching a vertex with the minimum cost in predicting a new image from all the images of a compressed image set, this new photo was added as a child of the found vertex. Compared with the former method, that further search may bring higher coding efficiency for the latter algorithm.…”