Research And Implementation Of Plant Root Segmentation And Quantitative Analysis System Based On Image Processing

Root is a vital part of the plant. The growth status of root can reflect the region’sclimate and soil properties, and whether the root is flourishing or not will response thehealth of the whole plant. Measuring the root parameters is the basis for studying thegrowth status of root, and it is important to the development of agriculture, forestry andecological. Root is not easy to observe directly while it is hidden under the surface.Minirhizotron can get the root images from the soil directly without affecting thenormal growth of plants, and it brought great convenience to the study of plant roots.Therefore, developing a system that can process plant roots images and measure theroot parameters accurately is particularly important. How to extract the roots accuratelyfrom images is the key and foundation of future research.The article focuses on plant root image segmentation. With the help of imageprocessing technology, we designed and implemented root image process andquantitative analysis system on Matlab platform, which is based on characteristics ofroot image information for quantitative analysis, and the measurement results arepreserved for analyzing the growth status of roots. The main research contents in thisarticle are as follows:①At the respect of image segmentation, threshold method, region growingmethod, the segmentation method based on edge detection and method based on thespecific theory are introduced in this article. Using all kinds of algorithms on the rootsegmentation in Matlab environment and analyzes the advantages and disadvantages ofeach algorithm. Aimed at the problem of traditional methods which are difficult tosatisfy the requirements of this subject, the article proposed a kind of automaticsegmentation method in minirhizotron images. Firstly, extract centerlines ofminirhizotron images, then compute the directional information of each pixel incenterlines, finally use the directional local contrast method to detect pixels on eitherside of the centerline. Using the evaluation method of image segmentation algorithm,we evaluated and analyzed this new method.②In terms of quantitative analysis, according to the features of roots afterrefining pixel and the root image pixel linear relationship between size and actual size,analyze and calculate the average diameter of root system at all levels, length, surfacearea, volume and other parameters on the basis of the diameter of the grading range set by users.③Plant root image processing and quantitative analysis system was designed andimplemented in the Matlab R2012a environment. It mainly Includes the followingmodules: read, display and save the image, image segmentation, quantitative analysis,the analysis results show, the result data preservation and batch processing.By researching plant root image segmentation algorithm, we constructed the imageprocessing and quantitative analysis system, which can provide reliable rootmorphological parameters for researchers in the field of root system. It is of greatsignificance for researchers to master the growth status of roots.