Combining image gradient and fractal dimension for object recognition

Object recognition is very important for many vision applications. A new method combining image gradient and fractal dimension is presented for object recognition. Fractal dimension is utilized to measure the distribution of image gradients of pixels. Then, the descriptor based on fractal dimension and image gradient is constructed to describe the image content. Object recognition is performed in term of the similarities of the image descriptor. The experimental results on three data sets demonstrated that our method is efficient and effective for object recognition.