SYSTEM AND METHOD FOR OBJECT RECOGNITION

摘要

A system for object recognition includes an image/spectrum sensing device, to fetch an object image from a real object and sense spectra at sensing regions of the real object. A fetching module obtains a real-object image feature pattern for each ROI of the object image. An analyzing module for object image feature searches for a first candidate object from data bank and analyzes a correlation between the real object and candidate object. A fetching module for object spectrum feature obtains a real-object spectrum pattern for each ROI of the object image. An analyzing module for object spectrum feature is to search for a second candidate object from data bank and to further analyze a match level between the real-object and the second candidate object. A fusion module further analyzes information of image feature and spectrum feature find whether or not having a matched object as an identify object.

主权项

1. A system for object recognition, comprising:
an image/spectrum sensing device, to fetch an object image from a real object and sense spectra at a plurality of sensing regions of the real object with respect to the object image; a data bank of object image feature, storing a plurality of image feature patterns corresponding to various first object samples, wherein each of the object samples is identified by a patch and the patch is formed by an atom array, wherein each of the image feature patterns is a sparse code representing an image information of multiple featuring atoms distributed in corresponding one of the patches; a data bank of object spectrum feature, storing a plurality of spectrum feature patterns corresponding to various second object samples; a fetching module for object image feature, coupled to the data bank of object image feature to obtain a real-object image feature pattern for each region-of-interest (ROI) of the object image; an analyzing module for object image feature, to search for at least one first candidate object for each ROI from the image feature patterns based on the sparse codes of the patches and to analyze a correlation between the real-object image feature pattern and each of the at least one candidate image feature pattern and give a weight and output a first information about the at least one first candidate object; a fetching module for object spectrum feature, coupled to the data bank of object spectrum feature, to obtain a real-object spectrum pattern for each ROI of the object image; an analyzing module for object spectrum feature, to search for at least one second candidate object of each ROI from the spectrum feature patterns and to further analyze a match level of spectrum between the real-object spectrum pattern and each of the spectrum feature patterns of the at least one second candidate object and give each of the at least one second candidate object by a weight and output a second information about the at least one second candidate object; and a fusion module, to analyze the first information and the second information to determine whether or not at least one identified object exits in the first information and the second information to identify the real object.