Development of Real-Time Vision Application on the Off Shelf Embedded Platforms

Abstract

Computer vision includes acquiring, processing, analyzing and understanding digital images from the real world in order to produce numerical or symbolic infor- mation. Embedded vision is a method of implementing computer vision algorithms on embedded systems. In the trend of miniaturization and smart devices, many industries are still working with large, costly PC based system. Objective of project is to develop an embedded vision system by integrating camera with single board computers. It provides advantage of small, low cost, light weight and less power con- suming system. One dimensional barcodes are found in retail, pharma, automobile and many more industries for automatic product identification to avoid any human error and compactly provides information about the product. The key contribution of this project is development of real time and automatic barcode localization and identification algorithm which is necessary to scan quick and accurate barcode de- coding from barcode images. The real time application of 1D barcodes detection, angle correction and decoding algorithm is in python. The simulation result is fur- ther communicated serially. The barcode decoding algorithms are implemented on single board computers like Raspberry Pi 2B and Odroid C2 Board. The compu- tational complexity in terms of execution time is compared and results show that embedded vision system are computationally efficient as PC and can replace the traditional PC based system. GUI also designed using Qt Designer and python as per user requirement.