Research Of Plant Leaf Modeling Based On Phenotypic Characteristics

With the rapid development and wide application of computer technology, researchers ofmany disciplines study with the help of computer. Virtual plant is an interdisciplinary science, itsimulates the morphology and growth of plant by means of computer in order to explore the lawsof plant growth and find the ways to improve crop yields. While the leaf is the mainphotosynthetic organ of plant and the growth rule of leaf plays an important role in lightutilization and dry matter accumulation. Mathematical description and realistic modeling of plantleaf are research hotspots of virtual plant. Due to the limitations of the theory and method, theleaf model constructed with geometric parameters only is the approximate description of naturalplant, but not realistic. Only combine with the leaf characteristics, can we construct the realisticleaf model.Getting the leaf characteristics by image processing can avoid the complexity of humanmeasurement, so the modeling approach based on the technology above is the research focus ofrelated fields. The images used in this paper were collected by biological experiments. Thispaper extracted the leaf characteristics by image processing, then constructed thethree-dimensional leaf model with the constraints of leaf characteristics and lighting area. Themain work done is as follows:1. Measure the leaf area accurately by directly processing colorful image.For the problem of using image processing to accurately measure the plant leaf area, thispaper proposes a new method that utilized the improved Snake model to extract the leaf contourand calculated the leaf area based on the leaf contour. First, this paper improved Snake model tomake it suitable for colorful image, on one hand, defined a color gradient in HSI color space asthe external energy function of snake; on the other hand, made the corners detected as thevertexes of the initial contour, and increased or reduced the vertexes by the algorithm. Second,the paper utilized the snake model to extract the leaf counter. At last, calculated the leaf areabased on the chain codes of leaf contour. The results prove that compared with Image J thisalgorithm is more accurate.2. Construct the accurate model of leaf contour.Aiming at the realistic modeling of plant leaf based on images, a novel algorithm of usingcontour feature points of plant image to interpolate cubic uniform B-spline is proposed. Firstly,the corners of the image were identified and extracted as feature points, then utilized the featurepoints to interpolate cubic uniform B-spline, finally new nodes were added by the improvedalgorithm of knot insertion according to the error to iterate until the required accuracy is met. Experimental results demonstrate that the new algorithm is superior compared to other methodsin the aspects of computation time and simulation effect. The vein is simulated by L-system, andthe vein is added to the leaf contour.3. Considering the effect of leaf area index on soybean production and based on the abovework, construct the three-dimensional model of soybean leaf.Aiming at structuring the realistic three-dimensional model of soybean leaf based on images,a method for three-dimensional modelling of leaf with the constraint of leaf area is proposed.Firstly, the feature points and leaf area were extracted form the image with the methods above.Then the feature points were interpolated by a bi-cubic uniform B-spline surface, the surfacegrids were deformed according to the change of leaf edge and the rolled leaf model wasconstructed. Lastly, calculated the maximum lighting area of each model separately and chosethe optimal model. The experimental results showed that the method could construct the realisticthree-dimensional model of soybean leaf. The method has certain biological significance andapplication value.