Abstract: A pattern recognizing method and apparatus are arranged to detect one or more objects with its own individuality belonging to the same category, such as a vehicle or a human's face, by using incremental signs in a manner to correspond with an apparent change caused by the posture variation of the object. For achieving the pattern detection corresponding with the apparent change caused by the posture variation of the object, the statistic quality of the incremental signs is extracted from a database having image data of the objects. The learning of a feature vector composed by using the quality makes it possible to design the most approximate identifier for detecting a pattern.