| 摘要 |
A multiple font optical character recognition reader system for capturing character patterns by pixel, quantizing and sequencing the data representing multiple pixels, enhancing the character patterns represented by the data, locating characters within the enhanced data, and recognizing the characters. Data representing individual pixels is quantized into white, gray and black levels. During enhancement, gray pixel data is selectively changed to black or white while some black is changed to white, both operations depending on the white, gray or black levels in adjacent pixels. The enhanced data is scanned for a multiplicity of simultaneous conditions which define when a matrix of the data includes a character pattern which is optimally positioned within the matrix. The data matrix containing the character pattern is then sequentially compared to sets of data individually representing templates of potential characters. When optimal matching and threshold conditions are satisfied, a character is classified as recognized. |
