National Central University

Industrial Technology Research Institute

Liang-Chien Chen

Jiann-Yeou Rau

Tee-Ann Teo

Chien-Liang Liu

Jer-Jiunn Chen

Wen-Chi Chang

Chih-Yuan Huang

Kuo-Hsin Hsiao

Chia-Hwa Lin

Ta-Ko Chen

Yun-Jou Lin

Yi-Chen Shao

Chi-Chung Lau

Shin-Hui Li

Jin-King Liu

A method for searching a building roof facet and reconstructing a roof structure line, in which the searching is performed automatically and without limitation of how slope of the roof facet, and the building structure line is constructed through aerial imagery. At first, lidar point clouds on the roof are extracted to compose a roof facet by using coplanarity analysis, and the roof is differentiated to a possible flat roof and a pitched roof. An optimal roof facet is obtained by analyzing lidar point clouds to overcome the low pitched facet issue. A relationship of a roof facet on a 2-dimensional space is analyzed to ascertain an area of a roof structure line. An initial boundary is generated. Line detection is performed on the images and a roof structure line segment is composed. All the structure line segments are used to reconstructing a 3-dimensional building pattern in object space.

Method for searching a roof facet and constructing a building roof structure line

The present invention combines an existing photogrammetric mapping system for building-model reconstruction. Topographic maps and models are produced simultaneously. The models can be edited along with aerial images in an online environment. Thus, efficiency and accuracy for building-model reconstruction are increased.

Method of online building-model reconstruction using photogrammetric mapping system

A method for true-orthoimage color correction is provided. Aerial images and digital elevation models (DEMs) are used for balancing colors in orthoimages or true-orthoimages. Seam lines between images are also smoothed. Thus, color distinction between images is rectified and orthoimage quality is greatly enhanced.

Method of orthoimage color correction using multiple aerial images

Lidar point clouds and multi-spectral aerial images are integrated for change detection of building models. This reduces errors owing to ground areas and vegetation areas. Manifold change types are detected with low cost, low inaccuracy and high efficiency.

Method of change detection for building models

A method and system for target detecting, editing and rebuilding by 3D image is provided, which comprises an inputting and picking unit, a training and detecting unit, a displaying and editing unit and a rebuilding unit. The inputting and picking unit receives a digital image and a LiDAR data and picks up a first parameter to form a 3D image. The training and detecting unit selects a target, picks up a second parameter therefrom, calculates the second parameter to generate a threshold and detects the target areas in the 3D image according to the threshold. The displaying and editing unit sets a quick selecting tool according to the threshold and edits the detecting result. The rebuilding unit sets a buffer area surrounding the target, picks up a third parameter therefrom and calculates the original shape of the target by the Surface Fitting method according to the third parameter.

Target detecting, editing and rebuilding method and system by 3D image

A high-resolution remote sensing image is processed through a true ortho-rectification. A totally new idea of H-buffer is provided to store heights of objects on the ground. The ortho-rectification includes a hidden analysis, a hidden detection and a hidden compensation. The process uses polygon patch of a building or a roadway as process unit. In the end, seam lines after hidden compensation are smoothed. The whole process time is reduced in calculations, assures quality of the orthoimage, and meets a requirement of a high-accuracy and high-resolution digital mapping.

Method of ortho-rectification for high-resolution remote sensing image

The present invention provides coefficient for a strip-based image taken by a satellite. By using the coefficient, the absolute error of the image can be made smaller than two pixels. Therefore, the accuracy of the image is high. The present invention can be applied to any pushbroom scanning satellite.

Method of generating positioning coefficients for strip-based satellite image

A semi-automatic reconstruction method of 3-D building models using building outline segments is introduced. The core technology of the present invention is called the 'Split-Merge-Shape' algorithm. The Split and Merge processes sequentially reconstruct the topology between any roof-edges of the buildings and then reform them as enclosed regions. The Shape process uses height information and consecutive-coplanar analysis to determine the shapes and heights of the roofs. After generating polyhedral building models, prismatic building models can also be generated by using a semi-automatic procedure. An existing digital topographic map of buildings can be directly used to reconstruct their 3-D models without any excess stereo-measurements. In addition to cost reduction, high efficiency, high quality, and minimization of manual operations, the integration of photogrammetric mapping with 3-D building modeling in one procedure is possible, which is the most cost-effective approach for 3-D mapping.

Growing community of inventors

Tao-Yuan, Taiwan

Liang-Chien Chen