Neural network segmentation and recognition of text data on engineering documents

The implementation of a novel neural network strategy for the segmentation of text information in the noisy, cluttered, and unconstrained environment typical of an engineering document is described. Additionally, the use of a neural network classifier for segmented mixed handprint and machine print characters is discussed. The results of the application of the system to the problem of extraction and recognition of the text information of water-network maps is presented.<
>