Research On Three-Dimensional Reconstruction And Visualization Of Maize Roots Based On Magnetic Resonance Imariniz And VTK

Roots are the nutritive organs of plants, mainly responsible for absorption of water and inorganic nutrients, organic nutrients storage and synthesis. The spatial dimension for water and nutrients absorption and the ability to compete with roots of other plants are basically determined by the directional distribution, length and surface area of a root system. Therefore it is important to study the roots systems.However, it’s very complicated to study plant roots. The mutual covering among different root systems along with the opaqueness of soil, make it hard to discover the dynamic growth circumstances and the interaction between roots and soil. The more conventional methods are always time-consuming based on destructive sampling. First separate the roots from soil, and then run the factitious measurements. The entire process was a true inefficiency and usually followed with massive inaccuracies. It was impossible to detect and measure the roots system accurately and in situ.The MRI (Magnetic Resonance Imaging) can perform an immediate and accurate analysis of materials construction without damaging the inside, which makes it a widely used technology in modern days.In this paper, with Maize roots as our research objects, an MRI-based non-destructive analysis method is reported. The research carried out a3D reconstruction of corn root systems with a combination of MRI and VTK technology and studied the external factors that affect the imaging quality. Also, in order to upgrade the quality, accuracy and efficiency of imaging, and study the results of3D reconstruction, a WINDOWS OS-based3D visualization system of plant roots was developed using Visual C++programing software.This paper includes:1) Studied the influences of external factors (as different soil media species, water content, etc.) by analyzing the MRI images of Maize Roots. After comparing the MRI images, the proper parameter settings (Thk/sp is2mm/lmm; TR is1800ms; TE is10.2ms) were decided. Research on the MRI images of Maize roots growing in different kinds of soil mediums proved that soil species have no significant influences on MRI image quality. And roots MRI images under different water contents of soil were captured for research, the result indicates that when water contents of soil was between5%and20%, it has no great influence on MRI imaging quality.2) Studied the abilities of different image processing algorithms by performing different pretreatments (image filterings, image interpolations and image segmentations, etc.) in order to find the best pretreatments combination. And the best pretreatment combination was decided to provide the optimized data for the following3D visualization process and this will upgrade quality of the3D reconstruction model. The experiment results showed that Contourlet wave filtering had a much higher SNR ratio than other two methods, which means Contourlet algorithm is more suitable for roots MRI image smoothing. The MRI images after filtering treatment were analyzed by threshold segmentation or region growing method. And threshold segmentation method can separate the root systems from the background more precisely than region growing method.3) Studied and developed the algorithms for3D reconstruction of Maize roots. Ray casting method was used to reconstruct the3D model of maize roots MRI images together with the algorithms provided by VTK toolkit. And by improving the tri-linear interpolation method, discreting the projected light and stopping the light in advance, the rendering speed of ray casting method became31.9%higher than before without loss on image quality.4) Established a MRI-based3D visualization system for plant roots based on all the previous algorithms. This system was programed by VTK and VC++language on windows platform. All the inaccuracies of the morphological parameters between reconstruction models and actual roots were below3%which proved this system having great reliability. So it can be concluded that this system can realize non-destructive observation and3D reconstruction of plant roos in situ on normal PCs with outstanding visualization quality.