Eyeglass Frame Identification Using Pixel Measurement Method and k-NN Algorithm

Abstract

Eyeglasses have a variety of types and shapes recently. The shape of the eyeglasses frames are rectangular, square, oval, pilot, round, geometric, and wrap. This study proposed an approach them to recognize the shape of eyeglasses. The digital image becomes an important part of this research. Eyeglasses image is taken from IP Camera and other sources (internet). The image should be processed into grayscale, then convert it to the binary image to get the height and width of the eyeglasses. The height and width were used to perform feature extraction. It generates 6 attributes, 3 ratios of glasses height and 3 ratios of eyeglasses width. That six attributes are classified by the k-NN algorithm. Based on the tests performed the accuracy reaches around 58% - 71%