Shuhe Zhao

Segmentation of multispectral high-resolution satellite imagery using log Gabor filters

Image segmentation has been recognized as a valuable approach that performs a region-based rather than a pixel-based analysis of high-resolution satellite imagery. A scheme for segmenting the multispectral IKONOS image based on frequency-domain filtering is presented. The frequency spectrum of typical landscape objects is analysed first. The spectrum curves are comparable in logarithmic coordinates rather than in Cartesian coordinates; therefore the Gabor filters are superseded by log Gabor filters to extract the multiscale texture features from panchromatic band. Edge features then are calculated from the pan-sharpened multispectral bands based on the vector field model. Finally, the texture-marked watershed segmentation algorithm is implemented and the segmentation accuracy is assessed. The experimental results show that the developed scheme generated an effective tool for automatic segmentation of multispectral high-resolution satellite imagery and suppressing the over-segmentation problem of watershed transform.

Multispectral IKONOS image segmentation based on texture marker-controlled watershed algorithm

Segmentation has already been recognized as a valuable and complementary approach that performs a region-based rather than a point-based evaluation of high-resolution remotely sensed data. An approach to segmentation of multispectral IKONOS image based on texture marker-controlled watershed transform is presented. Primarily the texture and edge features are extracted from the response of log Gabor filtering. The texture features are obtained from the amplitude response, and phase congruency is introduced to detect invariant edge features. Then a method for multispectral IKONOS image segmentation based on band feature combination is demonstrated. After that an algorithm to combining texture with edge features is presented and used to implement the marker-controlled watershed segmentation. Finally empirical discrepancy is calculated to evaluate the segmentation results. It shows that the precision of right segmentation rate is up to 75% to 85%.

Feature detection from IKONOS pan imagery based on phase congruency

Phase Congruency is introduced as a frequency-domain based method to detect features from high-resolution remotely sensed imagery. Three types of objects were selected from the IKONOS pan imagery in Nanjing, i.e. paddy, road, and workshop objects. The Phase Congruency feature images were obtained by applying Phase Congruency model to these images with 2 octave log Gabor wavelets filters over 5 scales and 6 orientations. The outputs of space-domain based detectors Sobel and Canny are also presented for comparing to Phase Congruency. It is then shown the results that the magnitude of Phase Congruency response is largely independent of image local illumination and contrast, and Phase Congruency marks the line with a single response, not two. It is followed by a set of results illustrating the effects of varying filter parameters and noise in the calculation of Phase Congruency. It is found that Phase Congruency can obtain more accurate localization than space-domain based detectors because it does not need low-pass filtering to restrain noise first. The results also show that the noise has been successfully ignored in the smooth regions of the image, unlike the Canny detector results fluctuate all over the image.

Object-based classification using LiDAR-derived metrics and QuickBird imagery

Due to the strengths and weaknesses of the airborne LIDAR data and QuickBird multispectral data, an improved classification method is presented for extracting vegetation information, roads, and buildings. A plot located in San Francisco was selected as the study site. Firstly, ground points were extracted from the LIDAR data and resampled to build DEM and DSM, and then derived nDSM by subtracting DEM from DSM. Secondly, the intensity information derived from LiDAR data was processed to be distributed evenly, and then generated an intensity clustering image, which classified LiDAR points into two basic clusters. Finally, add nDSM and intensity clustering images to QuickBird image as two extra bands, and then we can extract vegetation information, roads, and buildings using their height, intensity and spectral information. The results showed that the method combined airborne LIDAR-derived metrics and QuickBird multispectral data has higher classification accuracy. The proposed method in the paper could be applied to larger research area and other fields.

A novel classification method based on texture analysis using high-resolution SAR and optical data

Data fusion technique is an efficient way to benefit multi-source, multi-platform, and multi-angle remotely sensed information. Optical imagery and SAR (synthetic aperture radar) data are complementary in terms of capability of data acquisition and image characteristics. With their different capability and their unique information content respectively, fusion of high resolution SAR and optical multi-spectral imagery can improve the classification accuracy in land use. Texture information plays an important role for class discrimination especially in SAR imagery for its backscatter is sensitive to the type, orientation, homogeneity and spatial relationship of ground objects. In order to take full advantage of multi-source remotely sensed data and combine different features of them, this paper put forward a data fusion method for high spatial resolution remotely sensed data based on texture analysis. Texture features of high resolution SAR imagery were extracted using GLCM (Grey Level Co-occurrence Matrix) method. The texture features were detected in 0°, 45°, 90° and 135° four directions, and the moving window size of 3×3, 5×5, to 31×31, 41×41, 51×51, and 61×61 were tested to analyze the influences among them. The selected texture features were added with SAR data to make classification next. Both the two imagery were classified using an object-based and rule-based approach. Then, a decision level fusion was implemented and the accuracy of classification result was improved from 78.7% and 83.0% to 88.8%.

An improved plane fitting based filtering algorithm for airborne lidar data

Due to the strengths and weaknesses of the existing filtering methods, an improved plane fitting algorithm is presented for filtering of airborne LIDAR data in the paper. Firstly, resampled LIDAR data were segmented using region growing method, and the largest connected region was selected as the initial ground area. Secondly, non-ground points in the initial ground area were removed by a slope threshold, which is suitable for the study area. Finally, the ground points were interpolated using Kriging method and the filtering results were gained. The dataset provided by International Society for Photogrammetry and Remote Sensing (ISPRS) was selected as the test data. The influences of different slope thresholds on the filtering results were given after filtering experiments. The result shows that the optimal slope threshold in this area was 0.5 and the overall error rate was only 4.29%. Contrast to the traditional filtering method based on plane fitting, the proposed method is more simple and practical.

Data organization appproach to spatio-temporal GIS in campus real estate

Driving by a happened event, entities vary from one state to another. Based on the rule, this paper analyzed the relations between events of entities and its states, and made an improvement on base state with amendments model. The improved model is named as multi base state with amendments model. The key idea of this method is to build more than one historical base state according to the frequency of event happens and the amount of data updates. And for the state between every historical base state, we merely stored the changed part but did not re-store the unchanged part. It overcomes the weakness of snapshot method which leads a great deal of redundant data, and also overcomes the drawback of base state with amendments method which will need a great amount of complex computation when historical state is rebuild. This model has been successfully applied to organize the spatio-temporal data of GIS in campus real estate information system. It is very convenient to rebuild house historical state.

Evaluation on wetland classification in Yancheng natural reserve, China using HJ-1 data

Remote sensing of wetlands is one of important aspect in application and research of remote sensing. This paper studied the HJ-1 CCD data (30m/pixel) applied to wetland resource analysis of the Yancheng national natural reserve. And we gave evaluation on ability of the HJ-1 CCD data in wetland classification. Firstly, we gave the spectral analysis of the typical wetland types. The extraction experiments were carried out by supervised and unsupervised classification methods. Finally, we compared the classification results with those gained from Landsat-5 TM data (June 2007). It indicates that the HJ-1 CCD data have good recognition performance.

Entropy-based Texture Analysis and Feature Extraction of Urban Street Trees in the Spatial Frequency Domain

A method of texture analysis and feature extraction of urban street trees in spatial frequency domain is described in this paper. The QUICKBIRD image of Nanjing acquired in July, 2007 was considered. The image was first transformed by 2-D discrete Fourier transform. Then the energy of the component in spatial frequency was calculated. Entropy in a region of 7x7 window was considered to evaluate the energy distribution of the image. A Gabor filter was designed to extract texture features of street trees by using the radius and angel information of the entropy image. The precision of the segmentation result is 79.96%. Odd Gabor filter was designed to detect the edge of street trees, and the experimental result is excellent.

Extraction of enclosure culture area from SPOT-5 image based on texture feature

The east Taihu lake region is characterized by high-density and large areas of enclosure culture area which tend to cause eutrophication of the lake and worsen the quality of its water. This paper takes an area (380×380) of the east Taihu Lake from image as an example and discusses the extraction method of combing texture feature of high resolution image with spectrum information. Firstly, we choose the best combination bands of 1, 3, 4 according to the principles of the maximal entropy combination and OIF index. After applying algorithm of different bands and principal component analysis (PCA) transformation, we realize dimensional reduction and data compression. Subsequently, textures of the first principal component image are analyzed using Gray Level Co-occurrence Matrices (GLCM) getting statistic Eigen values of contrast, entropy and mean. The mean Eigen value is fixed as an optimal index and a appropriate conditional thresholds of extraction are determined. Finally, decision trees are established realizing the extraction of enclosure culture area. Combining the spectrum information with the spatial texture feature, we obtain a satisfied extracted result and provide a technical reference for a wide-spread survey of the enclosure culture area.