Research On Image Segmentation Algorithm Of Rice Surface Root Based On MATLAB

The root system of rice is not only an important organ supporting plants to absorb nutrients,but also closely related to its surrounding environment.In recent years,the research of rice root system has been deepened.Studies have shown that the morphological characteristics of rice roots are closely related to the quality,resistance and yield of rice,and even the growth and development of rice plants.In the field of the quantitative research on the morphological characteristics of rice root system,there is great of significance to improve the mode of farmland management,breed of rice varieties and genetic improvement.In recent years,with the rapid development of phenotype technology,the growth situation of rice root was measured and analyzed in the method of image processing technology,combined with fertilization,irrigation,light,temperature and other environmental monitoring technology,image processing technology has become a new technological means for rice breeding and functional genomics research,and the root of image segmentation technology is one of the important basis for the subsequent phenotypic group analysis.This paper deals with image segmentation of rice root system,Combined with some experience in image segmentation,especially in the fields of medicine and fingerprint recognition.The segmentation experiments on the original image of the surface root system of rice have been carried out to verify the feasibility of the multiple segmentation algorithms.First,the rice seedlings were planted in the unilateral transparent root chamber.The roots were filled with soil with nutrient solution,and there was a certain angle of inclination with the ground to ensure that the root system could closely adhere to the transparent glass wall of the root chamber.The root chamber was placed in the outdoor environment,and the root image was collected regularly under natural light.The rice root images are collected and segmented by a variety of segmentation algorithms in MATLAB2016 image processing softwareThis paper mainly discusses three kinds of segmentation algorithms,which are the main-branch connection algorithm,the local threshold segmentation algorithm based on the morphological characteristics and the adaptive threshold segmentation based on the morphological characteristics.Based on the three algorithms,a comprehensive segmentation algorithm is proposed.Finally,the experimental results show that,although a large number of root details are retained,the structure of the main branch connection algorithm is slightly chaotic and the burr phenomenon is serious,The overall segmentation effect is poor;Local threshold segmentation algorithm and adaptive threshold algorithm have their own advantages and disadvantages.The advantages and disadvantages of these two algorithms are complementary.The local threshold segmentation algorithm based on morphological features can retain more details of the root,but the contour fracture is more serious;adaptive threshold segmentation algorithm is improved on the continuity of the image,burr phenomenon has also been suppressed,fine roots,however,cannot be observed.If these two algorithms can be combined together,a more desirable segmentation results can be obtained,which will provide important data basis for the following study of phenotypic group of rice root system.In the quality evaluation of segmentation results,the evaluation index of the comprehensive segmentation algorithm is better than the other algorithms,indicating the feasibility of the comprehensive post segmentation algorithm.