Dynamic Trees: Learning to Model Outdoor Scenes

Abstract

Abstract. This paper considers the dynamic tree (DT) model, first introduced in [1]. A dynamic tree specifies a prior over structures of trees, each of which is a forest of one or more tree-structured belief networks (TSBN). In the literature standard tree-structured belief network models have been found to produce "blocky" segmentations when naturally occurring boundaries within an image did not coincide with those of the subtrees in the fixed structure of the network. Dynamic trees have a flexible architecture which allows the structure to vary to create configurations where the subtree and image boundaries align, and experimentation with the model has shown significant improvements. Here we derive an EM-style update based upon mean field inference for learning the parameters of the dynamic tree model and apply it to a database of images of outdoor scenes where all of its parameters are learned. DTs are seen to offer significant improvement in performance over the fixed-architecture TSBN and in a coding comparison the DT achieves 0.294 bits per pixel (bpp) compression compared to 0.378 bpp for lossless JPEG on images of 7 colours.

Multiple Exemplar-Based Facial Image Retrieval Using Independent Component Analysis

“…Our system retrieved different images of the first, with and without scarf in images 1, 6, 14, 24. We find images of the first brown-haired female in 2, 4,5,8,13,15,17,19, and 21 with/without accessories like sunglasses, etc. We also find images of the second brown-haired female in 2, 7, 9, 10, 11, and 22 with various expressions.…”

“…From these clusters, DYNDEX then searches for objects (top -nearest neighbors) that are closest to the query object. Adams and Williams [19] explored the dynamic tree model for learning outdoor scenes using the expectation maximization (EM) algorithm. Viola and Jones [20] provided a computationally cheaper model for face image retrieval where the notion of an "integral image" was introduced.…”