You Hongjian

3D Reconstruction of Building Based on High-Resolution SAR and Optical Images

In high-resolution SAR image, the building can form the layerover and shade. Basing on the imaging equation of SAR image, we can extract the height, width and location information of the building. The high-resolution optical image can reflect the roof shape and location of the building. But the information in optical image is so rich that it can produce heavy error when extracting the roof of the building with some automatic methods. In order to extract the roof accurately, we attempt to fuse the SAR image with the optical image. Because the buildings layerover can reflect the buildings location, by the imaging equation of SAR image and that of optical image, we use the layerover of building in SAR image to confirm the approximate location of the buildings roof in optical image. Thus, we obtain a small window that includes the roof of the building, and use the ISODATA method to classify the small window to get the roof class. In the end, we use the height, width, location and roof information to reconstruct the building. The result proves that me method is a good idea.

Fast rectifying airborne infrared scanning image based on GPS and INS

You Hongjian, Shao Yun, Li Shukai (Laboratory of Remote Sensing Information Sciences, Institute of remote Sensing Applications, Chinese Academy of Sciences, Beijing 100101) Abstract: Airborne infrared scanning image is often distorted due to the change of the airplane’s attitude and position. Therefore, ground control points must be used to rectify the raw infrared image if traditional approach is adopted. A new method is tested in this paper to rectify scanning image fast based on GPS and INS. First the positions of some pixels in a scanning line are calculated, and then these pixels are used as ground control points to rectify the whole scanning line. The gray interpolation and strips mosaic are also discussed in this paper. The data of flight tests processed shows this method is feasible. Keyword: GPS, INS, rectify image, mosaic

Building extraction from DSM acquired by airborne 3D image

Segmentation and edge regulation are studied deeply to extract buildings from DSM data produced in this paper. Building segmentation is the first step to extract buildings, and a new segmentation method-adaptive iterative segmentation considering ratio mean square-is proposed to extract the contour of buildings effectively. A sub-image (such as 50×50 pixels) of the image is processed in sequence, the average gray level and its ratio mean square are calculated first, then threshold of the sub-image is selected by using iterative threshold segmentation. The current pixel is segmented according to the threshold, the average gray level and the ratio mean square of the sub-image. The edge points of the building are grouped according to the azimuth of neighbor points, and then the optimal azimuth of the points that belong to the same group can be calculated by using line interpolation.

Geo-referencing airborne interferometric SAR data

Synthetic aperture radar (SAR) plays an important role in the earth observation system owing to its all-weather capability. Having the ability of getting the altitude of ground and image simultaneously, interferometric SAR (InSAR) technology becomes used in many fields in recent years. Airborne InSAR system is also developed successfully in June 2004 in China and now many flight tests have been conducted. We can get the SAR backscatter energy and altitude of ground i.e. SAR image and DEM through unwrapping the phase. But the reference coordinates of DEM and SAR image are imaging geometry coordinates which use range number and azimuth number as reference. The imaging geometry coordinates must be changed into geographic coordinates if DEM and SAR image are used in practice, this procedure is called geo-referencing. Owing to its side looking property, the undulation of ground height influences SAR image severely, nevertheless the altitude of ground which is very useful to eliminate the influence of ground undulation can be generated through InSAR. It is very difficult to get the precise position and velocity of airplane in the past, but now we can get precisely position, velocity and attitude using GPS/INS integrated system with the development of inertial navigation system (INS) and the mature of GPS, especially differential carrier phase GPS. The position and orientation system(POS) of Applanix company can provide high accurate position, velocity and attitude, and its position accuracy can reach 0.05~0.30m and velocity 0.0005m/s by post processing using GPS and INS combined filter algorithm We apply Doppler and range equations to geo-reference the DEM and image of InSAR based on the position and velocity of platform which is determined by POS/AV system. The principle and accuracy of geo-referencing InSAR data are discussed in the paper and the test data are processed using this method. Error estimation and field accuracy checking are also presented and the result is satisfactory.

The effects of multi-path scattering on the SAR image of cylinder cavity

In this paper, the effects of multi-path scattering mechanisms on SAR image is deduced through range Doppler algorithms (RDA). The conclusion that the cloud phenomenon appeared due to the multi-path scattering mechanisms is detained. Through the analysis, the cloud caused by the multi-path in the down range is corresponding to focus mechanisms and the cloud appeared azimuth is non-focus. At last, the shooting and bouncing ray (SBR) technique is employed to calculate the scattering of the cylinder cavity and by combining with range Doppler algorithms (RDA), the SAR image of a cylinder cavity with underside closed is precisely given, considering the effects of the multi-path scattering mechanisms in different azimuth.

Research on statistical analysis of SAR images

The statistic distribution of SAR image plays an important role in SAR image processing, and it has a great influence on the results. In practical applications, the distribution is often specified by human through experience, but in fact the distribution changes with different pictures of different scenes. In this paper we first find a way to determine the form of the image distribution with Pearson System, and then we use the Chebyshev-Hermite polynomials and the statistical properties of SAR image to get the Edgeworth expansion. Finally, we practice the distribution simulation on four different types of scene (the ocean, the town, the mountain and the agricultural fields). The experimental results prove the efficiency and the accuracy of these two methods, and will lay a good foundation for the follow-up steps in image processing.

Cloud Reduction and Destripe of Space-borne Image Based on Wavelet Transform

It is difficult to process and analyze remote sensing owning to cloudy and strips in the image, so we propose a new method to reduce cloud in space-borne image with multi-temporal data based on wavelet transformation. The images are decomposed into low frequency information and high frequency information using Mallet algorithm. Normalized difference low frequency index between two temporal images with cloud is used to reduce the low frequency information because the clouds are low frequency information and they mainly appear in low frequency wavelet coefficients, and higher frequency information of image is also extracted from two high frequency information. The reconstructed image using reduced low frequency and extracted by high frequency wavelet transformation is a new image without cloud. Two real space-borne images with clouds are processed using the proposed method and the result demonstrates that the method is feasible. Stripes also appear in some space-borne images and they are high infrequency information, so they clearly appear in wavelet field, and we reduce high infrequency of strips in wavelet field. The striped image is decomposed into several layers using wavelet transformation, and the strips are appear clearly as lines in high frequency block in each layer. Projection histogram is used to reduce the lines in high frequency block The reduced higher frequency information ant low information are reconstructed into a new image without stripes. We use the wavelet transform method to reduce TM space-borne image with strips, and the processed result shows that wavelet transform is a better method.

Study of Nonlinear Magnification Method Based on Bezier Transformation

In this paper, a new nonlinear magnification method is proposed. The new method produces the effect of nonlinear magnification based on perspective projection and the Bezier curve is used as drop-off function. So the new method can enhance the local information and keep the global context. It can provide different representation by adjusting the distortion degree of nonlinear magnification especially. In this way, image interpretation and target recognition would be performed effectively.

A new method to locate high resolution satellite imagery without ground control points based on prediction

In this paper a new method is proposed to solve the problem of high resolution satellite imagery without GCPs by considering the consecutive imaging parameters. That is, the imaging parameters of consecutive imagery are calculated based on GCPs in order to set up the prediction formula, and then the imaging parameters of high resolution imagery can be forecasted. Thus the rigorous model is introduced to precisely locate imagery. QuickBird imagery is test and geo-referencing accuracy reaches 3-4 pixels.