METHOD FOR OPTIMIZING MASK USING GENETIC ALGORITHM FOR PATTERN RECOGNITION
申请公布号:JPH07121719(A)
申请号:JP19930285527
申请日期:1993.10.21
申请公布日期:1995.05.12
发明人:TAKEDA FUMIAKI;ONAMI SAIZOU;MONNO TAKASHI
分类号:G07D7/00;G06F15/18;G06N3/00;G06T1/00;G06T7/00;G07D7/12;G07D7/20
主分类号:G07D7/00
摘要:PURPOSE:To decide a mask optimum for identifying paper sheets by simulating three big features of crossing-over, selection and mutation in the mechanism of the evolution process of organisms with a computer when performing the pattern recognition of paper sheets. CONSTITUTION:The read signal of a line sensor 11 is time sequentially inputted to a data processing part 140, processed according to the input of a sample pulse SP, and stored in a memory 141 as a frame image. Only the part of a paper money is segmented from the stored data in the memory 141 by a paper money segmentation part 142 and the segmented image is inputted to a neuro circuit 150, and the identified result DR of the neuro circuit 150 is inputted to an identifying device control part 160 composed of a CPU, ROM and ROM, etc. In this case, since the neuro circuit 150 adopts the conceptions of crossing- over, selection and mutation in genetic engineering for providing the optimum mask pattern, the best mask pattern can be provided in a short time just like the organism takes the process of evolution.