Implementing precise geometric changes in a time-efficient manner to be employed in processing images in two and three dimensions
International Journal of Mechanical Engineering · 2023
In this study, we provide a new technique for determining the geometry of objects and their potential uses.The suggested technique utilises a hybrid system that draws on the Standard Hough Transform, Bresenham's raster scan algorithm, and the Eigenvalues of the covariance matrix.This study takes use of the fact that certain properties of geometric primitives like straight lines, circles, and ellipses are correlated with the size and shape of the covariance matrix's eigenvalues.Regardless of the segment's length or orientation, a line in the continuous domain will always have zero for its tiny Eigenvalues.If the shape of the item is a perfect circle, then the object's big and small Eigenvalues will be equal.If the object is an ellipse, the semiminor and semimajor axes are represented by the small and large Eigenvalues, respectively.Therefore, in this work, the geometric primitives in images can be extracted by using the large Eigenvalues and small Eigenvalues of a covariance matrix.The suggested methodology is used in several computer vision and pattern analysis programmes.Automatic traffic systems often employ a straight line detection method to locate the lanes.Coins are identified using circular object recognition, while human faces are segmented and detected using elliptical feature extraction.