Recognition method for low-contrast embedded billet characters
YU Zhe-jun · Journal of Wuhan Institute of Technology · 2012
Aimed at the low contrast feature and irregular edge of embedded billet character,the LoG operator based on the measure of edge was proposed.At first,the convolution masks with multiple angles were used to enhance the edge information of characters in each direction,and the closed contours were extracted through the detection of zero crossing points;Secondly,the area filling method was used to obtain the candidate connected domains;At last,the noise was eliminated according to the difference with character in morphology,therefore the information of characters was completely extracted.The experimental results show that the recognition algorithm for embedded character has a good recognition rate and stability,the information of characters is kept completely in the detection results and can be well used for the process of character recognition,which provides the key technology for the practical application of embedded billet characters.