Image Edge Detection Based On Wavelet NSCT Transform And Mathematical Morphology

Many algorithms in the field of image edge detection have been proposedcurrently,but the theory and algorithms presented there are still manydeficiencies,the edge of the object can not effectively detected in some cases.In this paper,aimed at advantages and disadvantages of wavelet transform andnonsubsampled Contourlet transform (NSCT),the two transformations werecombianed for image decomposition,NSCT replaced by the Laplace Laplasse pyramiddecomposition of wavelet transform,high and low frequency sub-image were getafter the decomposition,then the Modulus maxima and improved methods ofmathematical morphology separately applied on the edge detection,the two edges ofthe image fused together finally.the traditional edge detection method and theoretical knowledge will be usedwere studied,and a detailed analysis of the various advantages and disadvantages ofedge detection operator was made.On this basis,a detailed analysis of the wavelettransform and NSCT transform was made,as the wavelet transform lack ofdirectional,the image texture and contour can not effectively represented,althoughNSCT transform to meet the problem of directionality,but the transformation hasredundancy,so present a algorithm based on the combination of wavelet transform andNSCT transform,the combining algorithms has no redundancy,at the same time have aflexible sonband decomposition direction.Have the simulation experiment with thecombined transformation,theoretically and in practice have proved the effectivenessof this improvment.Secondly studied modulus maxima and basic theory of mathematicalmorphology,traditional morphological methods have a anti-noiseimprovement.Presented to the high and low frequency sub-image combining methoddecomposed,respectively by NSCT modulus maxima method and improvedmathematical morphology for image edge detection.The paper gives specificmethod,through simulation experiments and on the edge of the comparative analysisof test results,proved the effectiveness and practicality of the improvment of edgedetection method in this paper.