Digital identification system source code
no vote
Use code comments very detailed The first step: training network. Training using the training samples. (With or without training in this program because I have trained network parameters saved, the reader can be used to directly identify) Step two: identify. First, open the image (256-color); Thirdly, the normalization process, click on the "all at once"; and, finally, click on the "r" or use the menu to find the appropriate items for identification. Identification of results displayed on the screen, but also in the output to the file result.txt. The recognition rate is typically 90%. In addition, can also be a separate image open picture step by step work, but be aware that each step can only be implemented once again, but also executed sequentially. Specific steps as: "256-color bitmap to grayscale"-"gray image binarization"-the "de-noising"-"slant"-"split"-"standard size"-"c