PATTERN RECOGNITION IN DIGITAL IMAGES FOR IDENTIFICATION OF SPELLING LINES ADULTERATED BY USING K-MEANS ALGORITHM AND PCA

Resumo

Pattern recognition in digital images for the identification of lines in adulterated spelling using k-means algorithm and Principal Component Analysis (PCA). Color is used to classify objects of interest from the analysis and comparison of the numerical values of the data, pattern recognition and classification information from the statistical information extracted from patterns through the PCA. The images of the traits used in the study were obtained on equipment that reflects an ultraviolet spectrum which focuses on the trait examined and captured by the camera. The choice of images in the ultraviolet region is made experimentally due to its higher absorbance capability compared with images obtained in other regions of spectrum. Two classes of traits: adulterated and unadulterated. The features are those that receive tainted overlapping somewhere alien. Traces unadulterated do not receive the addition of any strange spot.