Title :
Geometric Constraints Histogram Descriptor Based on Curvature Mesh Graph for Image Retrieval
Author :
Xie, Yonghua ; OhEigeartaigh, Micheal
Author_Institution :
Inst. of Comput. Sci., Nanjing Univ. of Inf. Sci.&Technol., Nanjing, China
Abstract :
A new geometric constraints histogram descriptor (GCHD) based on curvature mesh graph for image retrieval is presented in this paper. Through this method, the edge and angle geometric constraints based on the curvature mesh graph are extracted firstly. Then the histogram algorithm is applied on the geometric constraints to obtain the histogram matrix for each curvature point. Finally, the histograms are statistically normalized to form a histogram descriptor with fixed dimension. Experimental results validate the robustness and invariance of the proposed descriptor.
Keywords :
geometry; graph theory; image retrieval; matrix algebra; statistical analysis; GCHD; curvature mesh graph; geometric constraints histogram descriptor; histogram matrix; image retrieval; statistical normalization; Computational efficiency; Computer science; Content based retrieval; Data mining; Feature extraction; Histograms; Image retrieval; Information retrieval; Robustness; Shape;
Conference_Titel :
Image and Signal Processing, 2009. CISP '09. 2nd International Congress on
Conference_Location :
Tianjin
Print_ISBN :
978-1-4244-4129-7
Electronic_ISBN :
978-1-4244-4131-0
DOI :
10.1109/CISP.2009.5304144