Object classification based on visual and extended features for video surveillance application

Object classification in computer vision is the task of classifying a given object in an image or video sequence to one of a set of predefined object categories. There are two main factors which affect the performance of object classification. These are image representation and classification. We investigate object classification using visual features such as Scale Invariant Feature Transform (SIFT), Speeded Up Robust Feature (SURF). These features are highly distinctive for textured objects while they ignore color information. But color is also very much important cue for object recognition. Considering this assumption, we use also color histogram as local features. For image representation, we use Bag of words (BoW) model and Naïve Bayes for classification. We extract visual and color descriptor at each interest point from image and combine them aiming to use as feature. The experimental result shows that our approach achieves 5% higher classification rate than only using visual descriptor.