A Research Of Biomimetic Recognition On Off-line Handwritten Chinese Characters

Human ocular apperceive is truly a robust character recognition system with fine bearable mistake ability which can adapts to all kinds of noises. Recently, much about handwritten Chinese characters recognition has been made. Biomimetic Pattern Recognition makes sense both in theory and practice when it is used in large classes pattern recognition, like Chinese characters recognition.The following are the major contents in the thesis:1, According to the relative changeless topological structure brought about compounding, connection and crossover of some basic Chinese character strokes, the process of human "cognizing" Chinese characters is analyzed. Double-weights elliptical neuron sequence that accords to basic strokes is defined to cover handwritten Chinese characters image. Basic strokes are obtained. Image of handwritten Chinese characters is transferred into geometric figure composed by fault tolerance Chinese characters strokes of different position. Represented basic stroke segment figure, the double weights elliptical neuron is obtained.2, The psychological process of stroke styles fault tolerance and identification of topological structure are analyzed in human "cognizing" Chinese characters. Six styles of Chinese characters strokes with fault tolerance are concluded. The composing rules of Chinese characters strokes neuron and the identifying rules of Chinese characters strokes neuron joint and intersection styles are defined.3, Imitated Chinese characters recognition process of human learning, memorizing, contrasting and judging, and aimed the best cover of Chinese characters strokes in Chinese characters figure, the characteristic knowledge data-base table of handwritten Chinese characters and sample Chinese characters are built. Fault tolerant matching rules are defined.Handwritten Chinese characters in SCUT-IRAC are tested. The experiment is proved to be close to human fault tolerance and veracity.