FIELD: computer vision.
SUBSTANCE: method comprises the following steps: obtaining a floor plan of a three-dimensional (3D) scene structure, reflecting the a priori structure of this scene; obtaining one or more scans of the scene image with the camera; finding the direction of gravity for one or more image scans; constructing an edge scan by filtering out points on one or more image scans which horizontal projections are statistical outliers; aligning the border scan with the floor plan; extracting structural elements from one or more image scans and optimizing the three-term cost function based on the geometric term, the floor term, and the wall term by pulling points on one or more image scans that correspond to the structural elements to the corresponding points on the floor plan to refine the camera pose in 3D scene.
EFFECT: increase of the accuracy of at least one scan of a 3D scene while automatically obtaining a complete scheme of this scene.
15 cl, 9 dwg, 5 tbl
Title | Year | Author | Number |
---|---|---|---|
VISUALIZATION OF RECONSTRUCTION OF 3D SCENE USING SEMANTIC REGULARIZATION OF NORMALS TSDF WHEN TRAINING NEURAL NETWORK | 2023 |
|
RU2825722C1 |
NEURAL DOT GRAPHIC | 2019 |
|
RU2729166C1 |
METHOD FOR VISUALIZING A 3D PORTRAIT OF A PERSON WITH ALTERED LIGHTING AND A COMPUTING DEVICE FOR IT | 2021 |
|
RU2757563C1 |
METHOD AND ELECTRONIC DEVICE FOR DETECTING THREE-DIMENSIONAL OBJECTS USING NEURAL NETWORKS | 2021 |
|
RU2776814C1 |
METHOD FOR PROVIDING COMPUTER VISION | 2022 |
|
RU2791587C1 |
SYSTEM FOR CONSTRUCTING THREE-DIMENSIONAL SPACE MODEL | 2023 |
|
RU2812950C1 |
DEVICE, METHOD AND SYSTEM FOR RECONSTRUCTING 3D-MODEL OF OBJECT | 2015 |
|
RU2642167C2 |
METHOD AND SYSTEM FOR REMOTE CLOTHING SELECTION | 2020 |
|
RU2805003C2 |
BODY MEASUREMENT | 2013 |
|
RU2635226C2 |
SYSTEM AND METHOD FOR RECONSTRUCTING SURFACE PROPERTIES AND DETERMINING THE POSITION OF DIGITAL 3D OBJECTS | 2022 |
|
RU2806729C1 |
Authors
Dates
2023-04-18—Published
2022-05-27—Filed