EN   |   中文



Satellite Imagery

the automatic orientation for a stereo / triplet of satellite images, quasi-epipolar resampling of stereo pair, stereo matching, generation of dense point clouds, DSM generation, Automatically transforming DSM to DTM, pansharpening, orthorectification, automatic mosaic (DSMs mosaic, mosaic of ortho-rectified image)


Performance features of MASI software

(1) Supporting batch processing. It can automatically generate batch file which include all commands of stereo matching of aerial images or UAV images in a whole image block. After the configuration file is set, only one push of button is needed to carry out stereo matching for all images in a whole block. When images are ortho-rectified, all aerial or UAV images in a whole block can be processed via a push of button. In addition, users can operate very flexibly. They can start multiple processing of stereo matching, which is a relatively computation extensive, according to the computational resources on the machine. They can also stop any command in the course of the whole procedure as long as they re-start the step next time.

(2) Supporting parallel computing in different processing levels. Not only concurrently processing of multiple tasks but also parallel processing in a single task is supported. Because the number of aerial and UAV images is large, concurrently processing of multiple tasks is more efficient. In MASI, tasks of stereo matching and ortho-rectification for aerial and UAV image can be concurrently processed. Because the size of satellite image is huge, parallel processing in a single task is more appropriate in terms of performance. In MASI, there are this type of parallel processing for stereo matching, pansharpening, ortho-rectification, automatic mosaic, DSM2DTM of satellite images.

(3) Supporting different operating systems, either Windows OS or Linux OS. MASI can run on the commonly used PC computer, and cluster computer usually located in data center or supercomputer center.

(4) The size of a single image processed is unlimited.

(5) Different versions, both version on PC and version on cluster computer, can be customized in the light of requirements.

Functional features of MASI software

(1) Supporting almost all aerial digital cameras with the form of frame, traditional optical aerial camera, and calibrated consuming camera. Almost all frame cameras are supported, such as DMC I/II/III, UCD, UCX, UltraCam Falcon, UltraCam Eagle, RCD 30, PhaseOne, Hasselblad, SWDC-2, SWDC-4, etc. The consuming cameras used in UAV, e.g., Nikon, Canon, Sony, are supported. Most commonly used satellite sensors are supported, such as Worldiew-1/2/3/4, GeoEye, Ikonos, ZY3-01/02, Pleiades 1A/1B, SPOT 6/7, etc.

(2) Supporting different rotation system depicting camera pose, i.e., omega-phi-kappa, phi-omega-kappa, and kappa-phi-omega as well as rotation matrix. It means that all formats of the orientation list exported from aerotriangulation software can be used in MASI.

(3) Several cutting-edge techniques of stereo matching. In MASI, there is not only the method which can maintain the shape of building edge very well by finding the point corresponding for each pixel, but also the method which is optimized for the mountain area with large difference of height. There several steps of post processing for generation of highly dens DSM, i.e., interpolation of none data area in the DSM, outliers (small spot of errors) removal, refining.

(4) Advanced algorithm transforming DSM to DTM automatically. The algorithm can remove most buildings and trees in DSM, and can select different level of terrain details in the light of the requirements. It can improve the efficiency of generating DTM from DSM.

(5) A block-regression (BR) based pansharpening algorithm with satisfying tradeoff between maintenance of color content and enhancement of spatial details. BR algorithm can achieve a maximum enhancement of its spatial details at the expense of a minimum spectral distortion for the multispectral images. In the algorithm, the bands of the multispectral image used can not be limited and the algorithm is specially optimized for the pansharpening of very high resolution satellite image. In addition, parallel processing of the algorithm is also optimized.

(6) In ortho-rectification, image is rectified, pixel by pixel, and the geometric error is small. User can use DSM generated by MASI to truly ortho-rectify images, or use DTM to ortho-rectify them. It addresses different requirements of users.

(7) Four standard products (DSM, DTM, orthomap, pansharpened image) and an added-value product (difference map of height) are generated after the raw image is processed by MASI software. At the same time, some intermediate products are also yielded in the course of processing: point clouds, RGB point clouds, epipolar image and disparity image. These intermediate products can used in production of DLG (digital line graph), browsing and displaying, and quality inspection.

Comparative advances

Compared to other competitors, the main technical innovations in MASI software come from two aspects:

(1) Advanced techniques of dense stereo matching able to find the point corresponding for each pixel, DSM and DTM generation, and pansharpening

(2) Incorporating parallel computing into remote sensing image processing, resulting in promotion of speed and efficiency.

A series of parallel techniques with different processing levels, which can be used both on multi-core computer and on cluster computer, are suitable for different steps of image processing: pre-processing, stereo matching, ortho-rectification, mosaic, DSM2DTM, etc.



  Skype: jinghuiyang_bj
  E-mail:  jhyang@vip.163.com

Copyright: VisionOnSky Co., Ltd   备案号:京ICP备17045596号