Price tag recognition using HSV color space

Abstract

Object and character recognition from images are widely used for example plate car number recognition, scanner and even in the autonomous car. Researchers had developed lot of algorithms and image processing techniques to extract the features in the image such as hue, saturation and value (HSV) color filtering. However, the application of optical character recognition and image processing is not limited to car plate recognition but it can be implemented to recognize the other application such as price tag recognition. This project employed HSV color model for filtering the color and several image processing techniques to get the Region of Interest (ROI) and Optical Character Recognition (OCR) to recognize the product name and price in the price tags. Three price tags from shopping malls namely A, G and T are selected and analyzed in Android system. The image of price tags is filtered by color and applied rectangle contour detection to find the bounded area of the price tags with background image. Then, the process continues with masking the specific part in the price tags to recognize the character. In this method, color and contour detection able to separate the ROI with background and almost get the desired result with detection accuracy of 92.3% for price tag G, 90.63% for price tag T and 59.4% for price tag A.