METHOD USING RECOGNITION INTEGRITY, RECOGNITION TASK-ORIENTATION AND ADAPTABILITY FOR THE CONVERSION OF COMPUTER CODES DURING THE PROCESS FOR RECOGNISING THEIR CORRESPONDING IMAGE FRAGMENTS

摘要

The present invention pertains to the field of electronics and relates to a method that essentially involves, based on integrity, task orientation and adaptability, the preliminary creation of a base of volume hypotheses and/or of a functional group of volume hypotheses and/or of a recognition volume scenario that comprises a functional group of hypotheses and/or a set of trajectories (sequences) of hypotheses displacement. During the recognition process from the scenario thus formed, the method involves displacing a series of hypotheses selected, e.g., according to the representation volumes of the computer codes of each hypothesis. The method further involves checking their execution in a successive, integral and task-oriented manner relative to the scenario created until the error level value of an hypothesis validity, during the recognition of each semantic unit, is equal to or lower than an acceptable error level value, otherwise the end of the scenario cannot be reached. During the recognition process, which is based on the adaptive use of the results from the execution of the previous recognition phases, the method involves modifying the recognition scenario in a cyclic manner according to the complexity and the number of adaptation systems used.