Zernike Descriptors Invariant Rectangular Research

A feature description of the shape of the image is an important researchdirection and content in the area of digital image processing, pattern recognitionand computer vision.Blur and affine deformation are introduced during theimage acquisition process by moves of device,wrong focus etc.In resentyears,Finding a way to descript and recognize the image has become animportant research topic.To solve the problem of affine deformation of the images,the affineinvariants of Zernike moment was constructed to eliminate the influence ofaffine transformation.Firstly,the XSR decomposition method has been adoptedto decomposition the affine matrix；Secondly,the method of normalization wasused on the matrixs had been decomposed to eliminate the impact of affineparameter;Finally,based on the related theory of Zernike moment,the affineinvariant of Zernike moment was obtained and the description ablity has beenproved by the invariance experiment.Combined blur and affine moment invariants by Zenike moment wasproposed for the problem of image containing both blur and affine deformation.In the first, the theory of centrally symmetric point-spread function is applied togain the blur moment invariant of Zernike;After that the combined invariants toblur and affine transformation using Zernike moment obtained by combining theblur Zernike moment invariant and the affine Zernike moment invariant.lastly,the performance of description of the proposed moment invariants ispresented by proved by the experiment of invariance experiment;under differentnoise conditions, the comparison experiment with combined blur and affinemoment invariant comfirmed the low noise-sensitivity of the moment invariant;in the case of containing noise,the method has a higher recognition rate iscertified though the experiment of recognition.