Detecting Word in Digital Images Using Two Dimensional Fractal and Deep Belief Network for Military Applications

The algorithms of automatic movement recognitions are mainly applied in military and monitoring purposes including movement recognition, face recognition, and automatic targeting. In the study, we have used image processing and neural network models. The image processing has widespread application in security and military uses. The deep neural networks are developed from typical neural networks for training in nonlinear transformations. In this research, we have proposed an algorithm for recognition of handwritten digits. To do so, after input of each image we have executed preprocessing and then feature extraction. The methods of GLCM and multi-fractal have also been employed for feature extraction. The PCA has also been applied to reduce the dimensions of the features. The data obtained from the feature extraction have been input for the deep neural network. An intelligent system has been designed to recognize handwritten digits for military purposes.