| 摘要 |
PURPOSE:To extract a candidate category name at a high speed and with high accuracy by calculating evaluated values only as to some of category names selected for each vector element by using a dictionary in which category names are sorted in the order of element value for each vector element of a standard feature vector. CONSTITUTION:A selection processing means 2 which includes a processor 5, a program memory 6, a general memory 7, etc., refers to a dictionary memory 1 for each element of an input feature vector. This memory 1 consists of a dictionary table and a dictionary control table and stores the dictionary wherein category names are sorted in the order of element value for each vector element, and standard feature patterns are read out of the dictionary table corresponding to some of category names selected for each vector element by referring to the control table and matched. This collated result is cumulated by an evaluated value integration processing means 3, candidate category names are outputted through a sorting processing means 4, and only some of the category names are processed to extract the candidate category names at a high speed and with high accuracy.
|
