Segmentation Of CT Slices For Plant Root In Situ

Root is an important organ for plant to absorb nutrient from the surrounding environment.It is more difficult to do research than the surface plant for the root's special growth environment and complex morphology.Actually,quickly and accurately to acquire the root in situ to do deep study is a big technical problem.In order to solve this problem,the agricultural imaging detection technique laboratory in South China agricultural university has made some progress and got those CT slice images about the plant root in situ by the technology of X-ray computer tomography.And then used the method of image processing to realize the in situ observation and quantitative measurement of plants.What's more,the image segmentation is the basis to reconstruct the 3D architecture and measure the plant root in situ,and also plays an important role in the study of nondestructive testing of the three-dimensional configuration of the plant root in situ.To do this study,this group has tried many image segmentation algorithms to solve different problem when they meet,such as threshold segmentation,region growth,multiple differential operator edge detection and so on.But for the irregular growth of the lateral root,which grows to the surface of the ground called the inverse growth in the system of 3D digitization and visualization,a stable and reliable solution has not been found.Inverse growth refers to the non-geostrophic growth phenomenon caused by the factors such as dielectric barrier and nutrient induction,and the growth direction in the spatial sequence image is contrary to or inconsistent with the growth direction of the main root.The main technical bottleneck of the inverse growth problem is that the current segmentation algorithm is to select a seed point for the ROI region(root region)of each slice,cluster the pixels in a four-neighborhood or eight-neighborhood search path,Point down(on)to the projection,so that the surrounding pixels search clustering,it will always limit the single slice,can not search the same slice of the sudden appearance of the same area is not the same branch,In this paper,an in-situ CT sequence image segmentation method based on threedimensional connected domain extraction is proposed in combination with the characteristics of plant in situ root CT image.The method can realize the search clustering of the upper slice pixels in the segmentation.This method can solve the problem of inverse growth in the CT segmentation of plant root system,and ensure the integrity of the target area.The main work of this paper includes:1.This paper analyzes the present situation of CT slices segmentation technology at home and abroad,and compares the current CT slices segmentation methods,including region-based image segmentation method,edge-based image segmentation and segmentation based on specific theory.And on this basis,according to the task requirements,put forward the idea of this algorithm.2.According to the spatial density distribution of plant roots and the characteristics of inverse growth region,a segmentation algorithm of CT slices for plant root in situ based on the three-dimensional connected domain extraction technique is proposed.Firstly,according to the distribution characteristics of the root density,all the ROI regions including the root region are obtained by using the three-dimensional threshold method.Then,according to the continuity of the root in the media space,the three-dimensional connected region extraction technology based on 26-neighborhood is used to get the root region and the impurity which are similar density to root separated.The segmentation not only obtains a single and pure three-dimensional root target region,but also solves the problem that can`t get the inverse growth root by the traditional segmentation algorithm,which ensures the integrity of the target area.3.Using MATLAB programming,in the WIN10 system to achieve the segmentation algorithm programming and experimental testing.261slices(485×485×261),350 slices(485×485×350),358slices(465×493×358),a total of 969 slices of32(bits)gray image.respectively,were obtained by X-ray CT imaging equipment.At the first,to get the three groups of plant root package to 3D data and do the preliminary threshold value process and get the data of pixel's gray value rang from 1100~3300?900~34004?1812~3560.On the basis of this,the three-dimensional connected domain algorithm proposed in this paper is used to segment the coarse image data,and the impurity voxels with similar root density are eliminated.By using the mathematical morphology closed calculation,"Hole" filled,get a complete root target area.4.From the qualitative and quantitative aspects,the segmentation algorithm proposed in this paper is applied to the segmentation of CT images of in situ roots.Firstly,the results of the algorithm are compared with the traditional segmentation method.The validity of the segmentation algorithm in solving the inverse growth branch problem is verified by two dimensions and two dimensions.Then,the accuracy of the algorithm is evaluated by using the quantitative indexes such as the overlap rate of the partitioned region and the total voxel ratio.The results show that the algorithm has the advantages of 90.88%,90.06% and 90.41%,respectively,which are higher than the traditional segmentation algorithm by 3.90%,5.91% and 56.43% respectively.The algorithm is divided into three groups The total voxel ratio of the root zone was 93.61%,90.23% and 91.09%,respectively,which was 5.20%,4.01% and 53.26% higher than that of the traditional segmentation method respectively.In conclusion,the basis of using the 3-connection area to do image segmentation algorithm in this paper that can find inverse growth root and solve the problem,which can not be solved in the tradition algorithms,it is true to get more precision of segmentation about the plant root's.