Submitted to IEEE Transactions on Image Processing
A novel primary object discovery (POD) algorithm, which uses reliable object proposals while exploiting the re- currence property of a primary object in a video sequence, is proposed in this work. First, we generate both color-based and motion-based object proposals in each frame, and extract the feature of each proposal using the random walk with restart simulation. Next, we estimate the foreground confidence for each proposal to remove unreliable proposals. By superposing the features of the remaining reliable proposals, we construct the primary object models. To this end, we develop the evolutionary primary object modeling (EPOM) technique, which exploits the recurrence property of the primary object. Then, using the primary object models, we choose the main proposal in each frame and find the location of the primary object by merging the main proposal with candidate proposals selectively. Finally, we refine the discovered bounding boxes by exploiting temporal correlations of the recurring primary object. Extensive exper- imental results demonstrate that the proposed POD algorithm significantly outperforms conventional algorithms.
Yeong Juh Koh and Chang-Su Kim, "Unsupervised Primary Object Discovery in Videos Based on Evolutionary Primary Object Modeling with Reliable Object Proposals," submitted to IEEE Transactions on Image Processing.