Applied Research In Image Segmentation Used By The Algorithm Of Watershed

Image segmentation is a important content in computer vision and pattern recognition, is a basic operation of target detection,feature extraction,target recognition and so on, is also a key step from image processing to image analysis.The algorithm of watershed which based on morphology is early proposed by Rafael C and Beucher to using in image segmentation.It is well known as quickly calculation and accurately positioning the edge of image.It is perfect by Luc Vincent and Pierre Soille who propose a process based on immersion to achieve the algorithm of watershed.A method for watershed image segmentation based on shape templates is proposed to avoid over-segmentation.First, morphological opening and closing is employed to smooth the original image, afterSmoothing,unirnportant details and noise which are often the causes of over-segmentation are removed.Secondly,in order to preserve the essential region contours,the standard watershed transform is used to segment the smoothed image.Finally, combined with the Normalized Product Related Gray Match Algorithm to use the shape templates to direct the region merging and figure-ground segmentation.this method can efficiently not only avoid over-segmentation,but also achieve experimental results demonstrate the merits of this method.A method for watershed image segmentation based on wavelet transform is proposed to avoid over-segmentation.First, a adapt threshold denoising method which based on wavelet used for multi-resolution analysis is employed to smooth the original image,after Smoothing, noise which are often the causes of over-segmentation are removed.Secondly,we design a marker-extracted approach to extract the regional minima related to the object from the reconstruction of wavelet.And then extracted markers are imposed on the original gradients as its minima,while all its intrinsic minima are suppressed.Fially, the watershed algoritnm is applied to the modified gradients by the markers to reduce effectively the over-segmentation.and achieve experimental results demonstrate the merits of this method.