Analyzing and exploring feature detectors in images

In recent years, computer vision is being applied extensively in industry solution. It allows obtain color, shape and texture information in different situation. But in some applications, resolution and frame rate could limit it due high computation cost. This paper proposes analyze the most recent methods to detect feature in images in order to know the limitation in terms of computational complexity. It allows knowing where and when this kind of method could be applied. The most used methods, SIFT and SURF, are explored. Computational complexity is obtained analytically and compared with experimental results obtained with standard implementation. The results show similarity between the complexities, with advantage to SURF, due constants size.