A Fast Algorithm Of Bessel-fourier Moment And Its Application In Image Signal Real Time Detection

As Cloud Computing developing, the requirements of transmitting and retrievingmultimedia information such as images and videos are growing. NowadaysContent-based image retrieval (CBIR) system has becoming a new research hotspot.As an image feature, Bessel-Fourier Moment(BFM)is a set of moments based onthe Bessel function of the first kind. And it has a good retrieval effect in CBIR system.Overlapped Rectangle-Based Image Representation(ORBIR) is based on anti-packingpattern representation model(NAM). ORBIR obtains high compactness and is capableof supporting fast image operations.Basing on ORBIR, a fast algorithm of computing BFM is presented in the paper.The algorithm transforms computing BFM for multi-value images to a set of BFMson binary images that only contain one rectangle. The transformation replacesintegrals on2-dimensions by discrete sampling, which has nothing to do with concreteimages and computer first, and therefore, reduces the computations complexity formoments generation.Test Bessel-Fourier Moment performance in image processing. First experimentis performed to testing image reconstruction from BFMs. On binary testing images,the test compared different performance with Zernike moments, OFM moments ofsame orders. With the test results, it could be easily concluded that BFMs imagereconstruction is superior to the other two moments. Second experiment compared thedifferences of computation time between computing BFMs directly and computingBFMs with ORBIR. From the theoretical analysis and the experimental results, it canbe safely concluded that the new algorithm of computing BFMs with ORBIR costonly approximate half time less than direct computing. Third experiment calculatedBFMs absolute volume of a binary image and its multi-angle rotated images. It couldbe concluded that BFMs have the ability for rotated images retrieval.