Digital Recognition of Weighing Instruments Based on Machine Vision

Abstract

After image graying, binarization, filtering, corrosion, expansion and other related pre-processing operations, the tilted digits are vertically corrected by searching for the smallest external rectangle of the digits, and then the images are digitally segmented. Finally, the digits are recognized by using BP neural network algorithm, which are all based on the VS Software Platform (Microsoft Visual Studio) of computer and OpenCV Machine Vision Library. BP neural network algorithm has higher accuracy and can recognize different forms of numbers compared with traditional digital recognition methods. However, the processing time of digital recognition is longer, the number of digital samples processed is larger, and the algorithm is more complex compared with the general algorithm. We collect 300 digital pictures for recognition and test, and the recognition accuracy is as high as 97% in the digital image acquisition and recognition system. Finally, experiment shows that the method performs well in recognition accuracy and anti-jamming.