A new profile identification method based on multi-scale energy feature

As one dimensional range profile is sensitive to direction, translation and intensity, it needs extra storage and big computation in target identification. To reduce storage and computation, a new multi-scale energy feature based on dyadic wavelet transform is proposed. As energy is not dependent on the translation of range profiles, the new feature has a translation invariable property. The energy of two neighbor scales is divided to eliminate the intensity sensitivity. The feature is also insensitive to direction variation. When the proposed feature was applied into real data for target identification, good result has been obtained, which is better than conventional methods. Therefore, the new method is an effective feature selection method for range profile identification.