Skip to Main Content
This paper presents a biologically inspired object recognition algorithm that is tolerant to two dimensional (2D) affine transformations such as scaling and translation in the image plane and three-dimensional (3D) transformations of an object such as illumination changes and rotation in depth. The algorithm achieves this goal by extracting object features using Gabor wavelets and self-organizing maps in a hierarchical manner. Object features are learned in an unsupervised way which is consistent with the feature learning process in the visual cortex. The algorithm is analyzed for robustness. A support vector machine (SVM) classifier is used to test the classification efficiency of the algorithm.