| Root is one of the important components of the plant and the main organ of growth and metabolism. Root architecture reflects the plant growth in a way, so it is very meaningful to understand plant growth and other applications by studying root architecture. But underground root is hard to be sampled, measured and observed accurately, especially research is a time consuming work and lacks of its method, what's more the sample is easy to be destroyed, which block the deep study of root.In order to attain the equivalent machine filter effect based on the grazing waste-filter system, model method is used to study root's architecture feature. A root visualization model is put forward based on computer vision, measurement and pattern recognition and perennial rye-grass (Lolium perenne) root 3-dimensional graphic is generated as well. The concept of equivalent-filter precision in machine filtrate is defined; meanwhile the equivalent-filter precision and filter efficiency are computed using the root visualization model, which provide a further theoretic approach to design and research grazing waste-filter system. On the basis of analyzing the J2EE architecture specification, a prototype is carried out, that research established the foundation of distributive approach of visualization model. The detail content is listed as follows:(1) The principle of the root visualization model based on computer vision, measurement and pattern recognitions is listed: L-System is used as the main framework, and the L-System generated rules are created by image pattern recognition method, namely the topological information. Image-based measurement is used to get geometrical information, and the geometrical parameter is random modified at random according to that, at last the 3-dimension root graphic is generated using the fractal dimension to depict the complexity of root and validate the model.(2) In the root visualization model based on computer vision, measurement and pattern recognition, the principle of L-System is discussed firstly, and the formalization of DOL is described, then a bracket L-System is inducted to describe the branch architecture. In the generation of L-System rule, because the root architecture has the self-similar finite fractal feature, the architecture of root image is analyzed. Considering that the complicated image is composed of simple subsidiary image(subsidiary root), the simplest subsidiary image is defined as the basal elements, from the aggregate of basal elements, the complicated image is described applying the grammar(rules of generating image), L-System rules can be made from these rules. In the paper, L-System rules is generated from the step ofselecting basal element, evolving grammar, and mapping the grammar based on the basal element to L-System grammar.(3) Measure the parameter based on the computer vision method. On the basis of image collection system design, the origin image of root is acquired. The image is thinned after the step of low layer process (contrast enhance, mid-value filter, sharpen, binary process), then the skeleton of root is derivated. After the above processes, we can get the string of Freeman chain code, root length, angle, the number of root, side-root's distance, and fractal dimension.(4) After generating the geometrical parameter, we should add randomicity to reflect the actual growth and create more realistic effect because in the real environment root architecture is affected by many factors. The parameter is modified by FBM, normal school and uniformity distribution in the model.(5) In the process of generating 3-dimension root model, considering the feature of reasonable data organization, data-oriented expression and extensibility in XML, a expression of L-System called LSML (L-System Markup Language) is introduced, and A 3-D graphic engine is designed and achieved.(6) The concept of equivalent-filter precision is defined and applied to the effect of machine filtrate of root in the eutrophic water. Then equivalent-filter precision and... |
PDF Full Text Request