DIGITAL IMAGE CLASSIFICATION BY INTEGRAL-IMAGE FEATURE EXTRACTION

In the paper, we Introduce a new method for Image Classifi-cation. Suggested method uses the Integral-Image for Extracting featurevectors. We use the feed-forward backpropagation neural network andtraining this network by 50% data. This trained network , correctly clas-sifed more than 62.75% of the 800 images. And when we use the 100%datas for training, we obtain Excellent result. in this case, trained network,correctly classi ed 100% of the 800 images.

کلیدواژه ها:

Neural Network ، Integral-Image ، Feature Extraction ، Image Classi cation