Pattern recognition is the most popular are in the modern world. Significant progress has made in English character and digits recognition technology. However, the performance of the existing system of English character and digits recognition can be increased. This study is suggested a novel method, zig-zag diagonal feature extraction method along with ANN to increase the recognition rate of the English character and digit recognition system. The proposed method of this method is tested on an official dataset of English characters (EMNIST) and English digits (MNIST), achieving a 99.8% accuracy and 94% accuracy, respectively.