基于半监督深度学习网络的水体分割方法

水体分割是遥感图像语义分割的一个重要应用方向,现有的深度学习语义分割方法,无法有效分割复杂的水体边界、难以将空间信息和语义信息统一。为了解决上述问题,提出一种针对水体分割的深度学习网络。网络引入了注意力感知机制,能够对特征图中重要的信息加权,提高了对水体边界轮廓的感知能力。为了解决网络上采样过程中语义信息和空间信息不平衡的问题,将不同尺度的特征图融合进行预测。采用半监督框架对网络进行训练,充分利用无标记图像的特征信息,增强了网络的分割性能。实验表明,提出的网络分割性能优于其他现有语义分割网络。

基于混合模型的多尺度水平集高分辨率SAR图像水体分割

针对高分辨率SAR图像水体分割,提出基于混合活动轮廓模型的多尺度水平集分割算法。首先利用ROEWA边缘指示函数加权的轮廓线长度能量项和区域信息项建立模型能量函数;然后引入距离正则项,避免水平集方法在演化求解时需要不断重新初始化的问题;最后采用多尺度处理策略,解决水平集分割算法收敛速度慢的缺点。实测高分辨率SAR图像实验结果表明,该算法可用于水体的精确分割。

G^0和Gamma联合建模的水平集水体分割

针对传统水平集水体分割方法无法适应高分辨率SAR影像统计分布特性的问题,在影像金字塔结合水平集理论的多尺度迭代处理框架下,提出一种G^0和Gamma联合建模的多尺度水平集高分辨率SAR影像水体分割方法。该方法在金字塔顶层,考虑到水平集初始化对分割效率和性能的影响,引入快速模糊C均值聚类初始化水平集函数,然后基于G^0和Gamma联合分布对SAR影像整体综合建模和构建全局能量泛函进行能量最小化迭代分割,更好地拟合了高分辨率SAR影像的统计分布特性。最后,将得到的分割结果反演至下一层金字塔的分割,使分割结果在多尺度迭代链中不断精化,最终实现水体的精确分割。采用1m分辨率TerraSAR影像进行算法分割性能的对比实验。实验表明,该方法分割精度较高,是一种较为实用的高分辨率SAR影像水体分割方法。

基于图像分割的样本标注工具的开发

随着现代科学技术的飞速发展 , 图像标注已成了在计算机视觉和图像检索等技术领域中极为重要的一部分。随着现代互联网上的图像数量急速增长,为实现我们对信息的高效利用与检索,获得高精度的标注结果,需要我们建立一个集图像标注和管理系统于一体的图像标注平台。本文主要研究了基于图像分割的样本标注工具的开发,该样本标注工具能够实现半自主型的标注,实现人力的更大解放,同时我们引入了水域提取模块,能够实现人们对于遥感图像中水域形状的动态掌控,在洪水多发的夏季用于洪水的预防。

A method of remote sensing image water segmentation based on adaptive morphological elliptical structuring elements

The use of visible and infrared remote sensing images to calculate the water area is an effective means to grasp the basic situation of water resources,and water segmentation is the premise of statistics.Generally,the edge features of the water in the remote sensing images are complex.When the traditional morphology is used for image segmentation,it is easy to change the image edge and affect the accuracy of image segmentation because the fixed structuring elements are used to perform morphological operations on the image.To segment water in the remote sensing image accurately,a remote sensing image water segmentation method based on adaptive morphological elliptical structuring elements is proposed.Firstly,the eigenvalue and eigenvector of the image are estimated by linear structure tensor,and the elliptical structuring elements are constructed by the eigenvalue and eigenvector.Then adaptive morphological operations are defined,combining the close operation to eliminate the influence of dark detail noise on water without overstretching the water edge,so that the water edge can be maintained more accurately.Finally,on this basis,the water area can be segmented by gray slice.The experimental results show that the proposed method has higher segmentation accuracy and the average segmentation error is less than 1.43%.

A Tongue-images Segmentation Method Based on Local Restoration and Watershed Algorithm

Objective: To propose a method to segment tongue- images efficiently, and extract tongue- body accurately and quickly. Methods: Firstly, a kind of color- images' pre- processing technique was used to solve tongue- surface reflection problem. Neighbouring and similar region's information was used to restore the region with tongue- surface reflection problem by replacement. Secondly, the restored image was transformed into a gray one, and then processed by mathematical morphological operation- dilation to get a closed- loop edge. The third technique used was watershed algorithm, which is an usual tool in image segmentation. 'Watershed' function of matlab software was used to complete this algorithm. After that, region- combination technique was used. Through measuring neighbourship and similarity of regions, a non- objective and non- background region was merged into one of its neighbouring regions. This step was repeated until only two regions, objective and background regions, were left in the image. At last, corresponding to the merged image, tongue- body image was got from the original image. Results: 316 images were randomly taken from the image library for experiments, and 299 images were correctly segmented, so, the successful ratio is 94.62%. On the other hand, average time of running this method was about 50 s under whole sampling environment. Conclusion: The method presented in this paper can segment a tongue- body image from its original one effectively, and thus laying a good foundation for the following analysis work.

一种新的高光谱图像中太阳耀斑去除方法

太阳耀斑掩盖了水体的真实物理特性,严重干扰了水下特征的获取。根据被耀斑污染的像素和邻近未污染像素间的辐射差值,获得大气组分的吸收特征,并调节大气辐射传输模型中的模型参量,进而得到大气透射率。分析了影响耀斑污染程度的主要因素,依据近红外波段离水辐亮度近似为零,求出水体受耀斑污染程度的空间分布,据此进行耀斑去除,得到未被耀斑污染的图像。结果表明,应用该方法提取离水辐亮度,去除耀斑的效果较好。

Numerical simulation of deposit in confluence zone of debris flow and mainstream

Abundant solid materials were formed as a result of landslide and collapse due to Wenchuan earthquake.The solid source around mountains would form a debris flow when appropriate rain condition occurs.Such a debris flow is structurally very large and strong,and the river flow can hardly wash away the deposit when the debris flow enters into the mainstream.As a result,the deposit on the river bed due to debris flow will cause a series of hazards.Based on the previous researches and relevant data,this paper simplified the interaction between debris flow and current of the main river,and adopted the finite element characteristic-based-split algorithm which is favorable to the stabilization of dealing with the convection.Finally,the numerical model of the confluence of debris flow deposit and main river was developed,and the deposit progress of the mega-debris flow from Wenjiagou in Mianyuan river was reproduced.Furthermore,the influence of the deposit on the flow route of the main river,and distribution of velocity and water depth were analyzed.The results showed that the simulation deposit terrain qualitatively agreed with the field data through comparison,including the deposit area and depth distribution.Furthermore,the improvement of the model in future was discussed.