Shape recognition on embedded system

Abstract

Vision is the most advanced of human senses, so it is not surprise that images contribute important role in human perception. Shape recognition is an analogous to computer vision, which is an important field nowadays. Applications of shape recognition found in many areas, such as, manufacturing, medicine, space exploration, defences, etc. This project proposed method of shape recognition to detect circle, triangle, square, and rectangle in an image. The proposed method threshold the three dimensional RGB image input image to a binary image, morphological closing to fill holes of foreground pixel in binary image, and the use of bounding box to calculate object metrics. The object metrics are then compared with predetermined values that characterize the particular shape of an object. The algorithm was first developed in MATLAB and then implemented on raspberry pi. The algorithm applied in raspberry pi is coded in Python and result obtained is shown on a 16 x 2 LCD.