A Correction Method For Extracting The Central Line Of Plant Root Branch Segment

How to non-destructively and automatically measure the three-dimensional phenotypic parameters of plant roots inside soil medium is the subject that scholars in related fields including root biology are working on.Which is used to clarify the correlation between plant genotype and phenotypic information,improve the ability of plants to absorb water and nutrients,and understand their relationship with the surrounding environment.With the development of X-ray imaging technology,researchers have tried X-ray computed tomography(XCT)technology to perform in-situ detection of roots in growing soils and other media.In this technical scheme,the construction of three-dimensional vector model is the key link for extracting the phenotypic parameters of plant roots from the segmented root sequence images and the three-dimensional root model,and is an important basis for the measurement and analysis of three-dimensional root phenotypic parameters.The construction of the root three-dimensional vector model begins with the accurate acquisition of the center line or curved skeleton of the three-dimensional root system.The method to obtain root center line or curve skeleton according to the data foundations can be divided into two types: cross-section-based and 3D-models-based extraction methods(including solid models,surface models,and point cloud models,etc.).For the extraction method based on the cross-section: first is to obtain the root data from each layer segmented CT sequence images;then calculate the centroids of each area as the skeleton point;at last according to the topological relation to connect those points and form the root center line.This method to extract the center line of a single-rooted or independent branch root has some advantages such as simple theory,good centrality,easy implementation,low complexity and high operating efficiency.But for the transition branch root,this method has a fatal flaw in the principle of the algorithm,which will lead to the emergence of "fusion" and "break" errors in the root`s center line.The fusion and fracture errors not only lead to the error and positional deviation of the topological relationship of the root line in the branch transition zone,but also directly lead to the error of the location of the branch start point and the topology between branches.Therefor,with the basis of advantage of the cross-section method to extract the root`s center line,aiming at it`s original reason defect or error that exists on the center line of the transition section of the root branch.This article starts with the accurate locating of the transition branch root,and conducts related researches how to avoid the misconvergence of the centerline and restore it`s topological relationship.And according to the transition zone proposed a method to split the root sections based on the concave point detected,according to the except transition zone`s skeleton points using the cubic curve interpolation to reconstruct center line in the transition zone,to ensure the correctness and accuracy of subsequent 3D vector model construction.The main work done in this paper is as follows:(1)Determination of research objectives and technical solutions.Through the understanding of the early stage of the research work and the analysis of a large number of literature reading,detailed understanding of the three-dimensional skeleton and the centerline extraction method from domestic research situation and the existing problems,clear will be based on the cross-section method root branch center line of the transition section of correction as the research target of this article,and proposed the overall scheme and technical route of research work.(2)Localization method of root branch transition section.By analyzing the characteristics of different root segment in the root area,put forward the use of residual root area convex hull size and roundness change characteristics as constraint conditions,determine the transition branch root segment positioning rules,and design the corresponding localization algorithm,and realized the root branch transition section of the clear positioning.(3)The separation method of the root region in the transition section based on the concave point detection.In branch root within the period of transition,in view of the CT image sequences of root cross section area and the characteristics of the outline of the changes proposed by concave point detection in root cross section area of the transition segment separation method.Firstly calculates the convex hull,root regions and tag corresponding to the root zone of contours between convex buns,computing the root zone outline to convex hull on the edge of the distance,the distance as far as the root zone of contour points as pits in the root of section contour.Then,the connection between the pits is separated as the boundary of the root section area,and the separation of the root section area in the transition section is realized.Finally,the centroid points of the respective regions are calculated respectively as the new branch core points,and then the correction of the center line of the branch transition root segment is realized.(4)The positioning interpolation curve method to correct centerline.After locating the CT sequence images in the branch transition root segment,a skeleton point cloud model was constructed to divide the skeleton point into three parts: the branch transition root segment,the branch transition root segment inner circle less than the threshold and the roundness greater than the threshold;The number of layers of the last skeleton point in the main root outside the branch transition root segment and the number of layers of the first skeleton point where the roundness in the branch transition root segment is smaller than the threshold value are used to determine the starting point of the branch;the bifurcation starting point and the branch are respectively The skeleton points outside the transition root segment are interpolated three times,and the skeleton points in the branch transition root segment are added as the points on the new branch to correct the centerline.(5)To realize the skeleton line correction algorithm for branch transition root segment and verify it.Using the Visual Studio 2013 editor on a Windows system,using the C++programming language,image analysis library VTK,OPENCV to implement the key algorithm proposed above,and using the algorithm to test the test root and simulation root to verify the two centerline corrections The effect of the algorithm.At the same time,the root angle was calculated by the algorithm,recut method and manual measurement,and the comparison analysis was performed to verify the accuracy of the positioning algorithm.In this paper,the root length of the branch transition is measured by hand,and the comparisonanalysis is used to verify the practicability of the midline correction algorithm.In summary,this article uses X-ray imaging,computer graphics and computer graphics and other scientific techniques to focus on the improvement of the topological relationship between the skeleton points of branch transitions in the skeleton model based on CT sequence images.The midline of the root segment was corrected to obtain a skeleton midline model that fits the growth characteristics of the plant to construct an accurate vector model for root phenotypic parameter measurement.