华 珊, 陈 研, 梁露焘, 任图生. 利用基于偏微分方程的图像滤波技术研究土壤孔隙结构[J]. 农业工程学报, 2014, 30(3): 78-85. DOI: 10.3969/j.issn.1002-6819.2014.03.011
    引用本文: 华 珊, 陈 研, 梁露焘, 任图生. 利用基于偏微分方程的图像滤波技术研究土壤孔隙结构[J]. 农业工程学报, 2014, 30(3): 78-85. DOI: 10.3969/j.issn.1002-6819.2014.03.011
    Hua Shan, Chen Yan, Liang Lutao, Ren Tusheng. Studying soil pore structure by using image filtering technology based on partial differential equation model[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2014, 30(3): 78-85. DOI: 10.3969/j.issn.1002-6819.2014.03.011
    Citation: Hua Shan, Chen Yan, Liang Lutao, Ren Tusheng. Studying soil pore structure by using image filtering technology based on partial differential equation model[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2014, 30(3): 78-85. DOI: 10.3969/j.issn.1002-6819.2014.03.011

    利用基于偏微分方程的图像滤波技术研究土壤孔隙结构

    Studying soil pore structure by using image filtering technology based on partial differential equation model

    • 摘要: 利用自主研发的土壤切片数字图像处理分析系统软件,采用3种滤波方式,即BMO滤波(boundary mean oscillation有界平均振荡模型)、PM滤波(Perona-Malik偏微分方程模型)和中值滤波,对免耕、翻耕和旋耕3种耕作方式下不同深度土壤切片的数字图像进行去噪增强,并对处理后的图像进行了孔隙特征参数提取与统计。在此基础上,探讨了几种图像滤波技术的效果以及耕作方式对不同土层土壤孔隙形态结构的影响,以期找到适用于土壤切片数字图像的滤波技术,为后期土壤孔隙信息的提取与统计提供工具。结果表明,采用BMO滤波后,从土壤切片数字图像得到的土壤孔隙度与试验结果最为接近;其中,孔径<5 μm的孔隙度显著大于其他图像滤波技术的结果,而5~50 μm与>50 μm孔径的孔隙度与其他图像滤波技术的结果没有显著差异;与免耕相比,翻耕与旋耕提高了表层土壤的孔隙度,增加了传导孔隙的比例。

       

      Abstract: Abstract: In this paper, we applied an original image filter called BMO (bounded mean oscillation) and two other common image filters, the PM filter (Perona-Malik PDE model), and the median filter, to remove the noise and estimate the porosity and pore size distribution of digital images of a soil thin section obtained from three contrasting tillage systems (conventional tillage, rotary tillage, and zero tillage) and soil layers. The BMO filter uses a new derivative format to detect the boundary characteristics of images, and eliminates the influence of image noise using the mathematical method of integral average. The results showed that the BMO filter performed better than the other two filters in detecting image edges and detail characteristics. Compared with the other two techniques, the BMO filter gave a significantly higher porosity for the soil pore with aperture < 5 μm, and the soil porosity obtained from the BMO filter was the closest to the experimental data. The PM filter produced higher porosities for the pores with aperture 5 to 50 μm and > 50 μm than the other two filters, but no significant differences were observed among them. It followed that the PM filter was a little better than the other two filters in analyzing pores with aperture >5 μm. There were no remarkable differences between the soil porosities with aperture > 50 μm obtained from the three image filters and experimental data. However, for soil pores with aperture < 50 μm, porosities obtained from the image filters and experimental data were significantly different. The main reason for that was that the two-dimensional soil slice could not completely represent a three-dimensional structure of undisturbed soil samples. Image resolution may also impact on identifying small soil pores. The advantage of a BMO filter on identifying small soil pores to some extent compensated the loss of small soil pores caused by a preparation of soil section and image resolution. Our results also indicated that the conventional and rotary tillage systems improved soil porosity and the ratio of transmission pores (pores with aperture > 50 μm) in a surface soil layer. Zero tillage decreased soil porosity due to the reduction of mechanical disturbance on a surface soil layer, and showed beneficial effects on increasing the ratio of storage pores (with apertures of 5 to 50 μm).

       

    /

    返回文章
    返回