Scale invariant feature extraction for identifying an object in the image using Moment invariants

摘要

Feature extraction is the first and foremost activity in object recognition and detection processing. It reduces the amount of data by representing the image in the form of distinctive, representative interest points. This paper deals with the extraction of global features from the pre-processed images. Geometric Moment invariant produces a set of seven normalized moment invariants that are invariant under shifting, scaling and rotation. Geometric Moment invariant is widely used to extract global features for pattern recognition due to its discrimination power and robustness. After the feature extraction is done the dimensionality of the feature is reduced using the concept of Principal Component Analysis. Finally, the reduced feature vector is used for the recognition of object using the Nearest Neighbor.