Researchand Implementationof Plant Root Image Separation And Quantitative Analysis Method

In the research of soil-plant-atmosphere continuum (SPAC), soil-plant system is animportant subsystem. Plant roots is a top priority in soil-plant system research. From abiological point of view, plant roots can directly reflect the status of the plant growth.So, soil and atmospheric environmental impact on plant growth can be expressed inplant roots. With the rapid development of modern computer technology, especiallydigital image processing technology, people transfer the research method fromtraditional manual processing to computer automation which produces a qualitative leap.However, due to few studies abroad with plant roots research, there are still manyproblems to be resolved. This thesis is carried out under the Science and Technologyproject called Plant root growth dynamics in situ non-destructive monitoring andanalysis system developed. And the main content is to process the required plant rootsand obtain the quantization parameter with computer and digital image processingtechnology. The purpose is to provide data for people’s study on root growth conditions.Root image separation is one critical step in root image processing. Given that rootseparation operations need to be done on the binary root image with clear contour, thefirst problem to be solved is to separate the root target from background. After analysisof the features in original root image, global threshold processing technology is used toseparate the color image and has achieved better resultsSecondly, considering the features of root image that primary and lateral rootsbranching, overlapping and distribution network, a separation method based onmorphology and distance transformation is proposed in the thesis. The author appliesmorphology for skeleton extraction and corners inflation, uses improved distancetransformation to require the impact zone of partitioned root images and takes theborder of impact zone as the separation line. Programming in MATLAB andexperimental results show that this proposed algorithm can solve the problem of rootsand branches crossing effectively.Besides that, to further demonstrate the separation effect, several commonly usedseparation algorithm is chosen to do contrastive analyze in theory and experimentalresults in the form of chart display. Finally, this thesis uses MATLAB to design and develop the plant root processingand analysis system. The system contains image reading, image segmentation, imageseparation, quantitative analysis, experimental results show and data storage.