具有鲁棒性的彩色图像检索方法

提出了一种鲁棒的图像检索方法,它能够在检索大规模图像库时实现基于内容的快速匹配.匹配的索引来源于重量化HSV颜色空间得到的灰度图(Z值图)的区域特征.由于Z值图由反映色彩的聚类区域组成,并且使用矩方法来表达特征,因此该检索方法是鲁棒的,具有2维平移、尺度、旋转不变性,同时可以避免逐点匹配产生的时间耗费.     

基于无人机图像的密集匹配方法研究

本文提出了一种基于FAST算法的无人机图像密集匹配,使用FAST算法提取两图像的特征点,并使用归一化互相关算法实现两特征点点集的匹配,然后根据初匹配的结果采用RANSAC算法鲁棒地估算出基本矩阵和单应矩阵;采用极线约束和单应约束通过归一化互相关关系稳健匹配出所有点。实验表明,该密集匹配方法结果令人满意。     

利用亮度排序的快速鲁棒特征描述与匹配算法

针对快速鲁棒特征SURF描述符匹配精度不高且对光照变化不具有鲁棒性的问题,提出利用亮度排序的快速鲁棒特征描述与匹配算法。该方法在SURF算法的基础上,对特征邻域像素的灰度值进行排序和分段。通过建立索引表对每段的像素进行表示形成描述子,再将每段的描述子串联形成特征描述符对影像进行匹配。实验表明,该算法较SURF算法匹配精度高,匹配可靠性方面提高74.7%,且对线性及非线性光照变化均具有较好的鲁棒性。     

基于内容的遥感图像数据库植被区域检索

随着遥感图像数量的急剧增加,基于内容的遥感图像检索系统已成为一个研究热点。该文在以综合区域匹配算法为图像相似度度量的前提下,提出了以空间绿颜色特征值降序排序的遥感图像数据库植被区域检索方法。首先,用综合区域匹配算法对图像进行相似度度量;然后,依据这些图像的空间绿颜色特征值对图像进行降序排序,从而得到含有植被区域这一高级语义特征的检索结果。实验表明:该方法将检索查准率提高了20%,检索效率明显提高,可满足用户的需求。     

基于内容的遥感图像数据库城市区域检索

对于大型遥感图像数据库,如何快速准确地检索到需要的图像数据,是一个关键问题。该文在以综合区域匹配算法为图像相似度度量标准的前提下,提出了根据平均高频信号强度降序为排序标准的遥感图像数据库城市区域检索方法。首先,用综合区域匹配算法对图像进行相似度度量;然后,依据这些图像的平均高频信号强度,按降序对这些图像进行重新排序从而得到含有城市区域这一高级语义特征的最终检索结果。实验表明:该方法将检索查准率提高了27%,而且检索效率高,可以满足用户需求。     

一种改进包模型的遥感图像检索方法

提出了一种改进的包模型图像检索方法,使用本底图像构建视觉词汇字典。在建库过程中,通过近似KD-Tree搜索本底图像局部不变特征与入库图像特征间的对应匹配关系,并记录该关系;图像检索时,搜索检索图像与本底图像特征的对应关系,进而计算与入库图像的关系。该方法仅保存本底图像的局部不变特征,可实现感兴趣区域的图像检索。实验结果表明,该方法在提高图像检索准确性的同时,可以减少所需存储局部不变特征的数量。     

聚合CNN特征的遥感图像检索

针对高分辨率遥感图像检索中手工特征难以准确描述图像的问题,提出聚合卷积神经网络(convolutional neural network,CNN)特征的方法来改进特征表达。首先,将预训练的CNN参数迁移到遥感图像,并针对不同尺寸的输入图像,提取表达局部信息的CNN特征;然后,对该CNN特征采用池化区域尺寸不同的均值池化和视觉词袋(bag of visual words,Bo VW) 2种聚合方法,分别得到池化特征和Bo VW特征;最后,将2种聚合特征用于遥感图像检索。实验结果表明:合理的输入图像尺寸能提高聚合特征的表达能力;当池化区域为特征图的60%～80%时,绝大多数池化特征的结果优于传统均值池化方法的结果;池化特征和Bo VW特征的最优平均归一化修改检索等级值比手工特征分别降低了27. 31%和21. 51%,因此,均值池化和Bo VW方法都能有效提高遥感图像的检索性能。     

一种分块的基础矩阵鲁棒计算方法

针对基础矩阵计算过程存在异常数据的问题,提出了一种新的匹配点分块的基础矩阵鲁棒计算方法。该方法首先对特征点集进行分块处理,然后以点到基线的距离为最优化准则获得较为准确的特征点匹配集,最后以8点算法获得结果值。实验表明,设计的算法切实可行,对特征点错误匹配所造成的误差影响有较大的压制效果。     

利用三值模板模糊匹配进行手指静脉识别

针对静脉特征图像中静脉边缘和末梢模糊造成匹配困难的问题,提出了一种基于三值模板图像的模糊匹配方法。该方法将提取出的静脉特征图像分割为三值图像,值1代表目标区,0.5代表模糊区,0代表背景区。三值图像用于模板匹配时,计算模板间非0值区域之间的平均距离作为相似分数,分数越低,则模板越相似。该方法匹配能力强,不要求点与点之间的绝对位置匹配,对模糊的静脉边缘和末梢匹配具有较强的鲁棒性。对实验中采集的456张近红外手指静脉图像进行匹配,正确识别率达99.46%,表明本文方法是可行的和有效的。     

融合相容粒理论的遥感图像检索

为了提高遥感图像检索的效率和准确性,提出了一种融合相容粒计算模型的遥感图像检索方法。首先,根据相容粒理论定义了区域相容粒、图像相容粒和区域相容粒信息表等相关概念,将遥感图像粒化;然后,计算出图像区域相容粒的相似度;最后,结合综合区域匹配算法,提出融合相容粒理论的遥感图像相似性度量算法,并利用IKONOS影像进行对比实验。实验结果表明,融合相容粒理论的检索算法能够提高遥感图像检索的查准率,与综合区域匹配算法相比,本文算法查准率提高了12.08%,基本满足用户需求。     

一种图像目视效果的改善方法

常规的图像目视效果改善方法会改变像素的灰度值,灰度值的改变会引起各种各样的图像特征计算发生不同程度的变化,进而影响到图像地物的识别和分类工作。提出利用Windows系统常用BMP图像格式的特点,通过调整颜色查找表来达到既实现图像目视效果改善、又不改变像素灰度值的目的。     

一种高分辨率遥感影像阴影去除方法

基于阴影属性提出了一种全自动彩色影像阴影去除算法.首先将影像变换到HSI空间,依据阴影区域亮度值低和饱和度高的特性,结合小区域去除和数学形态学处理,得到精确的阴影区域.然后,分别对I、H、S分量图上各个独立阴影区域与其邻近的非阴影区域进行匹配补偿,再反变换回RGB空间,完成阴影去除操作.实验结果表明,该方法能在不改变非阴影区域信息的情况下,有效地去除阴影的影响.     

基于Roberts梯度与HIS色彩空间的SAR图像伪彩色编码

针对SAR灰度影像信息不丰富,不利于解译识别的问题,提出了一种基于Roberts梯度与HIS色彩空间的SAR图像伪彩色编码算法。首先,计算SAR图像的梯度图像,并确定梯度阈值T;然后,比较像素梯度与阈值的大小,大于阈值的像素按照Roberts梯度方法编码,否则按照基于HIS色彩空间方法编码;最后,对两种编码结果进行合成,得到最终伪彩色SAR图像。利用TerraSAR-X影像数据进行了实验验证,结果表明:本文研究的方法均能达到很好的伪彩色编码效果。     

Production of high-resolution remote sensing images for navigation information infrastructures

This paper introduces the image fusion approach of multi-resolution analysis-based intensity modulation (MRAIM) to produce the high-resolution multi-spectral images from high-resolution panchromatic image and low-resolution multi-spectral images for navigation information infrastructure. The mathematical model of image fusion is derived according to the principle of remote sensing image formation. It shows that the pixel values of a high-resolution multi-spectral images are determined by the pixel values of the approximation of a high-resolution panchromatic image at the resolution level of low-resolution multi-spectral images, and in the pixel valae computation the M-band wavelet theory and the à trous algorithm are then used. In order to evaluate the MRAIM approach, an experiment has been carried out on the basis of the IKONOS 1 m panchromatic image and 4 m multi-spectral images. The result demonstrates that MRAIM image fusion approach gives promising fusion results and it can be used to produce the high-resolution remote sensing images required for navigation information infrastructures.     

Production of high-resolution remote sensing images for navigation information infrastructures

This paper introduces the image fusion approach of multi-resolution analysis-based intensity modulation (MRAIM) to produce the high-resolution multi-spectral images from high-resolution panchromatic image and low-resolution multi-spectral images for navigation information infrastructure. The mathematical model of image fusion is derived according to the principle of remote sensing image formation. It shows that the pixel values of a high-resolution multi-spectral images are determined by the pixel values of the approximation of a high-resolution panchromatic image at the resolution level of low-resolution multi-spectral images, and in the pixel valae computation the M-band wavelet theory and the d trous algorithm are then used. In order to evaluate the MRAIM approach, an experiment has been carried out on the basis of the IKONOS 1 m panchromatic image and 4 m multi-spectral images. The result demonstrates that MRAIM image fusion approach gives promising fusion results and it can be used to produce the high-resolution remote sensing images required for navigation information infrastructures.     

Comparison of Remote Sensing Image Fusion Strategies Adopted in HSV and IHS

It has been always a challenging task to keep an ideal balance of spectral and spatial resolution for merging panchromatic image and multispectral image. The mathematical theories such as color space transformation and Wavelet Packet Analysis are usually employed in information fusion area. Combining color space conversion with wavelet packet theory is a way of researching remote sensing image fusion algorithms further. In the paper, there are three existing image fusion strategies applied to the second layer of frequency bands decomposed by wavelet packet analysis in the HSV and the IHS (triangular coordinate) color space, respectively. Serial experiments demonstrate two core concepts. One is the effects of image fusion strategies based on region is super to those of fusion strategy based on pixel for the same color space; the other is the different performances are measured in the two color spaces. Specially, the space definition for image fused in the former color space is inferior to that in the latter color space; while the spectrum content for image fused in the former color space retains better than in the latter color space, when using the same fusion strategy in the two color space. As a result, application containing HSV space conversion can alleviate spectral deterioration, whereas fusion operation of IHS transformation can lift spatial definition.     

Cascaded image analysis for dynamic crack detection in material testing

Concrete probes in civil engineering material testing often show fissures or hairline-cracks. These cracks develop dynamically. Starting at a width of a few microns, they usually cannot be detected visually or in an image of a camera imaging the whole probe. Conventional image analysis techniques will detect fissures only if they show a width in the order of one pixel. To be able to detect and measure fissures with a width of a fraction of a pixel at an early stage of their development, a cascaded image analysis approach has been developed, implemented and tested. The basic idea of the approach is to detect discontinuities in dense surface deformation vector fields. These deformation vector fields between consecutive stereo image pairs, which are generated by cross correlation or least squares matching, show a precision in the order of 1/50 pixel. Hairline-cracks can be detected and measured by applying edge detection techniques such as a Sobel operator to the results of the image matching process. Cracks will show up as linear discontinuities in the deformation vector field and can be vectorized by edge chaining. In practical tests of the method, cracks with a width of 1/20 pixel could be detected, and their width could be determined at a precision of 1/50 pixel.     

Automatic Road Extraction from High Resolution Satellite Image using Adaptive Global Thresholding and Morphological Operations

Road network extraction from high resolution satellite images is one of the most important aspects. In the present paper, research experimentation is carried out in order to extract the roads from the high resolution satellite image using image segmentation methods. The segmentation technique is implemented using adaptive global thresholding and morphological operations. Global thresholding segments the image to fix the boundaries. To compute the appropriate threshold values several problems are also analyzed, for instance, the illumination conditions, the different type of pavement material, the presence of objects such as vegetation, vehicles, buildings etc. Image segmentation is performed using morphological approach implemented through dilation of similar boundaries and erosion of dissimilar and irrelevant boundaries decided on the basis of pixel characteristics. The roads are clearly identifiable in the final processed image, which is obtained by superimposing the segmented image over the original enhanced image. The experimental results proved that proposed approach can be used in reliable way for automatic detection of roads from high resolution satellite image. The results can be used in automated map preparation, detection of network in trajectory planning for unmanned aerial vehicles. It also has wide applications in navigation, computer vision as a predictor-corrector algorithm for estimating the road position to simulate dynamic process of road extraction. Although an expert can label road pixels from a given satellite image but this operation is prone to errors. Therefore, an automated system is required to detect the road network in a high resolution satellite image in a robust manner.     

高分辨率航带影像像控点的布设和基准约束测量

推扫式航摄仪所获取的带状影像,其每个像素都具有比较准确的WGS-84坐标。带状影像模型自成刚体,具有很强的稳定性,但其区域网模型精度在高程方面偏弱。受其自身航带影像模型刚性和区域网弱高程模型精度的影响,其像控点布设有其自身的特殊性。在获取具有多重空间基准约束的像控点的过程中,当在像控区域中有CORS基站、GPS控制网、水准控制网等多个空间基准类型可以利用时,可利用IGS站点和EGM2008数据快速验证这些基准的应用有效性,避免已有空间基准的潜在问题,这是保证像控点快速和精确获取的关键。     

An Approach to Single Image Automatic Orientation and Point Determination by Using Ortho-Images and a Dtm

This paper describes an approach to single image automatic orientation and point determination by using current ortho-images and a DTM, and the experience gained in its implementation. The procedure proposed automatically extracts and matches feature points in evenly distributed patches on aerial images and ortho-images. A large number of image measurements (up to several thousand) are obtained in this process and are included in a robust space resection to determine the orientation parameters of the aerial image. For point determination with a single image, a method is formalised which integrates the DTM interpolation into the space resection so that the 3D ground coordinates of the image points can be determined in a unified mathematical model. Tests and analyses of this method show that the large number of automatic image measurements relieves the requirement for complicated and precise feature extraction and matching methods. The ground points obtained from single image intersection have an accuracy of approximately 1 pixel in planimetry, which fulfils the requirement for ortho-image updating. The elevation accuracy is mainly dependent on the quality of the current DTM and the interpolation method applied to it.