International Science and Technology Journal

Identifying written numbers is one of the most important problems in the computer, which has made many researchers research this topic and create intelligent systems to identify them. In this paper we will create a smart system that recognizes the digitally typed figures from 1 to 5,this system focus on pattern averaging, which is a technique that controls the size and remodeling of images to be suitable for the network, also this system relies on the technique of canny edge detection, which is the technology of identifying the boundaries of the image, finally by using the network back propagation neural network BPNN we will identify the numbers entered, The paper Divided into two main parts. Part 1: Using Image Processing Techniques We will collect the digitally typed images from 1 to 5, written by volunteer students in the computer department of the Higher Institute of Science and Technology, Alhraba. Then by using image processing techniques, we convert images to grayscale and then to binary. Then by using canny edge detection we will recognize Edge Detection and then we will resize the image by using pattern averaging technique to become a 8 * 8 matrix. The second part: - We will use the back propagation neural network BPNN to identify the images numbers. The results obtained from the training and testing phase of the network showed that the percentage of recognition during the training phase was 100 % , while the percentage during the testing phase was 88.88%