Analysis Of Composite Material Image Using Varying-weight Distance

The analysis of composite material has become an active research field in recentyears and the development of the research is playing an important role in area ofmaterials manufacturing. And it is leading a great prospect of application. In thispaper, Discrete Distance is mainly used based on varying-weight distance to process,analyze and research the images of composite material.In this paper, a path-based distance is considered where local displacement costsvary both according to the displacement vector and with the travelled distance. Thesedistances and distance transform extend neighborhood sequence distances, chamferdistances and generalized distances based on Minkowski sum[6]. The correspondingdistance transform algorithm, which is similar in its form to classicalpropagation-based algorithms, has been proposed to compute a translated version of aneighborhood sequence distance map with a finite number of neighbors in order tomeet the forward scan condition for each of them[6]. This method is used to computean asymmetric generalized DT based on any number of neighborhoods havingforward scan condition. And a method is also introduced to transform the resultingtranslated distance map back into a regular, symmetrical NS-distance map.The histogram of the Opening Transform (OT) is very important for analyzingthe images of composite material. But using mathematical morphology to the imagesdirectly is very time-consuming especially when the structuring elements are verylarge. It is proposed to use the Granulometry Method to approximate the histogram ofthe opening transform from the distance transform. This varying-weight distancetransform is used to estimate the histogram of the opening function of the image inorder to get the useful properties of the image. Compared with using mathematicaloperators to the images directly, this method improves the efficiency greatly.Meanwhile, the histogram which is computed in our experiments is connected withthe Granularity of the images which is one kind of important textural features. The experimental results can be used to research on human visual perception and then wecan fit out the relationship between the texture feature and the human vision. In thispaper, we mainly focus on2D images of composite material, it can be extended to3Din the future.