Real - time system for shape extraction from an image

Abstract

This paper presents a system of “image sensor” for recognition and counting of shapes present in an image. Result information is provided as identified object, position relative to the lower left corner of the image and the number of objects detected in a certain type of image. In other words, the system functions as a sensor, the image is the entry and the items extracted from it represent the output data. Experiments aimed basic shapes: rectangle, triangle, circle, but the work can be extended to other more complex forms. The system was implemented on a Zynq SOC 7000 (Xilinx) and can be used in many applications involving pattern recognition in real time.