A homestead house surveying and mapping method based on handheld slam and semantic segmentation
By using handheld SLAM devices and semantic segmentation technology in rural homestead surveying, high-precision 3D models are generated, solving the problem of extracting house outlines in complex occlusion environments and achieving efficient and automated output of surveying results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHONGQING XINRONG LAND & HOUSING SURVEY TECH RES INST CO LTD
- Filing Date
- 2026-05-13
- Publication Date
- 2026-07-14
AI Technical Summary
In rural homestead surveying, traditional methods cannot efficiently acquire full-element data and automatically extract house outlines in complex occlusion environments. Existing technologies suffer from large cumulative errors and difficulties in automatically distinguishing house outlines due to the presence of numerous debris in point cloud data.
A handheld SLAM device was used to collect multi-source sensor data. Combined with a vision-inertial navigation tightly coupled algorithm and GNSS positioning, a global point cloud map with absolute coordinates was generated. A semantic segmentation model for residential land and houses was used for component-level classification, building outline features were extracted, and a three-dimensional model was constructed.
It achieves high-precision and automated building surveying in complex occlusion environments, eliminates cumulative errors, improves the efficiency of field data collection and indoor mapping, and generates three-dimensional visualization surveying results that meet real estate registration standards.
Smart Images

Figure CN122391883A_ABST