Among the research areas of artificial intelligence, there is image recognition through neural networks. Based on mathematical models of neurons, neural networks as Neocognitron, were precursors to the image recognition. Seeking performance computing, computational models of neurons have emerged, which would become the weightless neural networks. The WiSARD was one of the first weightless neural networks, and through your neurons based on random-access memories, brought pattern recognition performance. The NC-WISARD is a neural network based on the Neocognitron’s multilayered hierarchical network, on the the weightless WiSARD perceptron and also on its variation with non-supervised training methodology, the AUTO-WiSARD. The multilayered hierarchy enables the correct recognition, not only of previously trained patterns, but also any variation of size, position and distortions that the image may have. With the hability of extraction of patterns that are simple and relevant to the images, the NC-WiSARD can, like Neocognitron, through its hierarchical structure, recognize and assemble more complex patterns.