A computer vision software to automatically measure dimensions of flat surfaces in a photograph

Abstract

This project aimed to develop a fully functional program using computer vision to measure the area of flat surfaces of any shape in a photograph. A previous project has been conducted and a substantial amount of work have been done on objects and contours detection. A program was developed using Python and measurement of rectangular surfaces has been realized with accuracy above 90%. The OpenCV library was used as the primary tool for image processing. The current project was built on top of the previous project, new object detection methods along with other supporting functions have been developed. The capability of the program has been extended and any flat surfaces of any shape can be measured. The accuracy of the measurement has been increased to 97%. Lastly, a Graphical User Interface (GUI) have been developed using the PyQt framework. This allows users with no programming knowledge to operate the program. Test results of the program have also been discussed in this report. In the discussion, possible causes of some issues have been identified, and future work have been purposed as well.