One Simple Method of Positioning and Segmenting Dot Matrix Characters for Recognition

Abstract

Dot matrix characters are printed on the packaging boxes or plastic bags of food or other products by inkjet printers, recording the manufacturing date and product ID. This information is of much significance for product trace and quality control. As dot matrix characters are prone to printing errors due to printer vibration, it is essential to check whether the printed characters can be recognized or not after printing. Most OCR (optical character recognition) technologies only recognize continuous characters and very few are devised for dot matrix characters, i.e., discontinuous ones. The major difficulties of dot matrix characters recognition lie in segmentation and positioning due to their discontinuous dot structures. In the paper, one effective and efficient recognition package has been developed and one simple but valid method is proposed to accurately position and segment dot matrix characters and it can handle the characters of adhesions and discontinuities. Two popular segmentation methods are selected as benchmarks and the experimental results show that the proposed method has the best effects of segmentation and recognition.