Joint distribution and steerable filter based fast local feature description algorithm

We propose a new fast local feature description Algorithm, which called Multi-Resolution Wavelet Transform Descriptor (MRWD), in order to decrease the disadvantages of the scale invariant feature transform (SIFT), e.g. instable to illumination, slow speed and high dimensions ect. Firstly, we establish the joint distribution of normalized pixels´ intensity and distance to eliminate the impact of lighting transform. Then, we achieved a steerable filter using the multi-scale feature of Haar wavelet, to reduce the computational complexity and dimension. Finally, generate the MRWD descriptor. The algorithm not only reduces the instability caused by illumination change, but also maintains a good robustness on the scale and rotation transform compared to the traditional distribution based local feature description algorithm. Meanwhile, MRWD is faster than SIFT in computational speed, and reduces the dimensions.