Department of Computer Engineering
S E M I N A R
Image Classification Using Subgraph Histogram Representation
Computer Engineering Department
We describe an image representation that combines the representational power of graphs with the efficiency of the bag-of-words model. For each image in a data set, a graph is constructed from local image features of interest regions and their spatial relationships. First, local patches of interest are detected using maximally stable extremal regions obtained by gray level thresholding. Next, these patches are quantized to form a codebook of local information, and a graph is constructed by representing these patches as the graph nodes and connecting them with edges obtained using Voronoi tessellations. Then, each graph is represented with a histogram of subgraphs selected using a frequent subgraph mining algorithm in the whole data. Transforming of the graphs into a vector space enables statistical classification of images using support vector machines. Experiments using images cut from a large satellite scene show the effectiveness of the proposed representation in classification of complex types of scenes into eight high-level semantic classes.
DATE: 29 March, 2010, Monday @ 15:40
PLACE: EA 409