发明名称 Learning method and apparatus for pattern recognition
摘要 A method for information processing includes a learning process to generate a tree structured dictionary based on a plurality of patterns including a target object to be recognized. The method includes selecting a plurality of points from an input pattern based on a distribution of a probability that the target object to be recognized is present in the input pattern at each node of a tree structure generated in the learning process, and classifying the input pattern into a branch based on a value of a predetermined function that corresponds to values of the input pattern at selected plurality of points.
申请公布号 US9053393(B2) 申请公布日期 2015.06.09
申请号 US201113050872 申请日期 2011.03.17
申请人 Canon Kabushiki Kaisha 发明人 Yoshii Hiroto;Matsugu Masakazu;Mitarai Yusuke
分类号 G06K9/62;G06K9/68 主分类号 G06K9/62
代理机构 Canon USA Inc. IP Division 代理人 Canon USA Inc. IP Division
主权项 1. A method for generating a dictionary being used for recognizing a target object by classifying a plurality of learning patterns the method comprising: obtaining a plurality of learning patterns including a target object, wherein each learning pattern has a same number of pixels; generating a plurality of mask patterns which respectively indicates a region where the target object exists in each learning pattern; calculating probability that a pixel at a same position in each learning pattern is included in the region where the target object exists by referring to the plurality of mask patterns; selecting at least one pixel having relatively higher probability based on the calculated probability; and classifying each learning pattern based on the at least one selected pixel.
地址 Tokyo JP