A lot of work has been done on digit recognition since Yann Le Cun et al's work in the early 1990s. The MNIST database that was used to benchmark the classifier is for handwritten digits (see : http://yann.lecun.com/exdb/mnist/ ) but any of the pre-trained convolutional neural networks available for this problem should work pretty well on digits extracted from a digital display. The CNN approaches are the highest performing in terms of regocnition accuracy. There are many implementations in C++ on github. Not sure about Visual Basic. In terms of understanding the approach, I would suggest reading Le Cun et al's paper:
Perhaps you can create examples of all digits in the OpenCV pattern matching, then use it to retrieve the digit's positions and arrange them into the number.
hello..you can use the cropping instruction in matlab language to crop the numbers from the image then do anything on the image like zoom_in , zoom_out or display it . the first way to cropping by giving the location and the second way by cross the x-axis and y-axis.
Thanks for your reply. Actually, our requirement is little different. It's not simply displaying the number but I need to perform some calculations on the extracted values. So it need to be in integers.