Plants Recognition using Leaf Image Pattern Analysis

Abstract

The features including shape and color profile can be covered up in image processing while evaluating a pattern under test. In image processing, a pattern is transformed from its shape to feature vector. Feature vector may include its perimeter, color profile, radii, area, edge features, moments and key points on pattern etc. While working or extracting features of a pattern, it is very much required that the features are invariant with respect to its size, orientation and location. Size invariance can be achieved via centre of mass of the pattern. Orientation invariance is obtained by using orthogonal transformation of features And size invariency is achieved using the mean radius of the pattern under test. In the existing techniques of pattern recognition, the features are dependent upon size, orientation and location. Therefore, a pattern recognition system lacks the faithfulness and repeatability for the same pattern if are resized or oriented at different angles. This issue of feature normalization has been taken care of by normalizing the features using different techniques. Size normalization is achieved by using mean radius.