Application Of Morphology Watershed Transform In Wood Cell Image Segmentation

Along with the developing of wood science,it has become an important direction of wood research to analysis wood cell image by Computer Vision.Image segmentation is an important approach to analysis cell area,number,and percentage of cell wall,which are all important parameters to reflect wood's property.It is also the foundation of calculation and analysis in Microcosmic Form of wood cell.As a new subject,image processing technology has enjoyed rapid development and extensive application of the military,industrial,aerospace,and other aspects in a short period. As the foundation of image analysis and understanding,image segmentation is the most fundamental in the field of computer vision,but also one of the most difficult problems.The division result has directly impact on the performance of the follow-up algorithm.Because the diversity and complexity in image,until now,there is still not a completely universal segmentation method could satisfy the demand of segmentation correctly for all kinds of image. Therefore image segmentation technology has being the hotspot in image processing.This paper briefly introduces the principle and function of various representative segmentation algorithm.Compared several traditional edge detection algorithm in the application of wood cell image segmentation.And starting with the principle of typically traditional edge detection operator,divided microscopic image of wood cell,and compared differences among different classical operators.Analysis their advantages and disadvantages in the wood cell microscopic image segmentation According to the problems which traditional edge detection operator incapable of resolving,put watershed algorithm model into the segmentation of wood cell's microscopic image.This paper focuses on the watershed algorithm,and through a form of experimental results show that advantages and disadvantages of the method in analysis of wood cell's morphology image.According to the characteristics of coniferous wood's cell image and disadvantage of morphology watershed algorithm,this paper improved the image segmentation model of watershed algorithm.First preprocessed model based on the partial differential equations,and removed noise,at the same time highlights the border;After filtering,transformed gray level difference of the images.Used these transformed images as a reference image in segmentation. Because of removing the possibility of over segmentation,this method laid theoretical foundation of wood cell classification,and got more accurate outcomes of identification.