作者: K. Etemad , R. Chelappa
关键词: Enhanced Data Rates for GSM Evolution 、 Edge detection 、 Time delay neural network 、 Computer science 、 Artificial neural network 、 Probabilistic neural network 、 Artificial intelligence 、 Algorithm 、 Very-large-scale integration 、 Nonlinear system 、 Orientation (computer vision)
摘要: An approach to the edge detection problem based on nonlinear mapping and generalization capabilities of multilayer feed forward neural networks is proposed. The task broken into two parts, i.e., typical gray levels in primitive small image blocks (e.g., 3*3 windows) their corresponding most likely patterns using a simple network, combining this locally derived information (including presence, orientation strength edge) consistent way. Some experiments scheme are provided. suggested scheme, because its parallel structure, fast can be easily implemented analog VLSI hardware. >