Computer vision framework for object monitoring

Abstract

This paper presents the use of Visual C++ integrated with OpenCV library as a framework for video analysis applications. Conventionally, applications are developed and control using textual-based Win32 console. In this work, however, the library can be applied in Windows form that helps users to visually control algorithms implemented in applications. The framework, written in OOP style, tailors the existing library into three main class methods; initializing, running and ending. Streaming data can be captured from USB cameras, IP cameras and video files. Implementing computer vision algorithms include object counting and object monitoring is demonstrated. Results show that algorithms work well and they can be developed in short time.