Image processing and object classification

摘要

A method for classifying objects from one or more images comprising generating a trained classification process and using the trained classification process to classify objects in the images. Generating the trained classification process can include extracting features from one or more training images and clustering the features into one or more groups of features termed visual words; storing data for each of the visual words, including color and texture information, as descriptor vectors; and generating a vocabulary tree to store clusters of visual words with common characteristics. Using the trained classification process to classify objects can include extracting features from the images and clustering the features into groups of features termed visual words; searching the vocabulary tree to determine the closest matching clusters of visual words; and classifying objects based on the closest matching clusters of visual words in the vocabulary tree.

主权项

1. A method for classifying objects from one or more images comprising the steps of:
generating a trained classification process; and using the trained classification process to classify objects in said one or more images; wherein generating the trained classification process comprises the steps of: extracting features from one or more training images and clustering said features into one or more groups of features termed visual words; storing data for each of said visual words, including colour and texture information, as descriptor vectors; clustering the descriptor vectors into a plurality of clusters termed codewords; generating a plurality of candidate vocabulary trees using a plurality of the codewords, each candidate vocabulary tree being generated by repeatedly clustering that plurality of codewords; determining, for each candidate vocabulary tree, a score value indicative of how well that candidate vocabulary tree classifies as itself each of the plurality of codewords from which that candidate vocabulary tree was generated; and selecting, from the plurality of candidate vocabulary trees, based on the score values, a vocabulary tree, the selected vocabulary tree storing clusters of visual words with common characteristics; and wherein using the trained classification process to classify objects in said one or more images comprises the steps of: extracting features from said one or more images and clustering said features into groups of features termed visual words; searching the selected vocabulary tree to determine the closest matching clusters of visual words; and classifying objects based on the closest matching clusters of visual words in the selected vocabulary tree.