| The root is an important organ for plants to absorb nutrient and water. This abilityis closely relevant to its architecture and distribution. The study of root is one of mostimportant subject of plant nutrition, plant physiology, etc. And the analysis of rootmorphology is important content in root research. The calculation and analysis of thebasic parameters of root morphology and root architecture parameters have a veryimportant significance for the identification of plant growth situation, as well as toprevent plants from diseases and insect. Root’s shape is quite irregular, so the geometricmeasurement of it has always been a very difficult task to plant researchers. This paperfocuses on plant roots scanned image. Through a combination of computer visiontechnology, it studies proper image processing for plant root scanned image and imageprocessing technology included and also puts forwards segmentation algorithm toabstract root image object, an algorithm of replenishment of image thinning skeleton,the method of calculation of root’s whole characteristic parameters and branches’characteristic parameters and the method of identifying root and lateral root. It stillconstructs a root characteristics analysis protosystem based on computer vision. Themain contents of this paper are:1. Study how to properly use mature image processing techniques to fit thescanned image of roots and their applicability improvement. Firstly, this paper turnsroot image into a grayscale image, then selects median filtering for image smooth anddenoising through experimental comparison and uses top-hat transform to even theimage background. Secondly, it puts forwards a proper segmentation method for theroot scanned image and uses closed operation to reconnect the disconnected branches.Finally, it uses a fast parallel algorithm for image thinning and designs an algorithm forskeleton’s replenishment.2. Study the extraction method of characteristic parameters of the plant roots basedon scanned image. This paper refers the current modeling method of calculation of thewhole root’s characteristics parameters. While obtaining the whole characteristicparameters, it put forwards the extraction method of each branch’s characteristic parameters (including the length of each branch, average radius, projected area, surfacearea and volume), as well as the angle between the respective branch.3. Study the method of the identification of main root and lateral root. This paperanalyzes the characteristics of main root, then puts forwards a Dijkstra-basedoptimization algorithm to quickly identify main root based on the trend of main rootcurve, and on this basis, it identifies lateral root of the plant roots further.4. Structure and implement an analysis protosystem to quick obtain the rootcharacteristic parameters. Firstly this paper states the main function, key technologiesand the overall system architecture of this system, then analyzes functional requirementand performance requirement. Secondly, it states the design of the class diagram andmain function of each class. Finally it designs a friendly software interface. |