i/rtabmap
1
0
Fork 0
mirror of https://github.com/introlab/rtabmap.git synced 2026-09-02 09:30:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

New feature: Depth confidence (#1520)

Browse Source 
* New feature: Depth confidence

* iOS app updated to save depth confidence, added util2d::depthBleedingFiltering function

* Updated tools to show/extract depth confidence

* Android: moved smoothing in post-processing, fixed confidence registration, added depth bleeding error option.

* Fixed warning

* removed debug log

* Added new feature types, fixed rendering when exporting texture >4096 (#1469), added depth bleeding filter option to iOS

* fixed some warnings, android: added bleeding error option

* CI: try updating ros2 key

* added sudo

* antoher test

* bump ios app version
This commit is contained in:
matlabbe
2025-06-01 14:14:29 -07:00
committed by GitHub
parent 90d195237f
commit 6d4e8a4173
51 changed files with 2368 additions and 1006 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
app/ios/Settings.bundle/Mapping.plist
View File

@@ -58,7 +58,7 @@
<key>Type</key>
<string>PSGroupSpecifier</string>
<key>FooterText</key>
<string>Choosing between high and low depth confidence depends on the application. If precise geometry is critical, prioritize high depth confidence even if it means fewer points. If you need a dense point cloud for analysis, even in areas with weak LiDAR signal, a lower depth confidence might be suitable at the cost of highly interpolated points (that could give a wavy look of some surfaces). Important: the points with too low confidence are not saved in the db file, so we cannot recover or filter these points in post-processing.</string>
<string>Choosing between high and low depth confidence depends on the application. If precise geometry is critical, prioritize high depth confidence even if it means fewer points. If you need a dense point cloud for analysis, even in areas with weak LiDAR signal, a lower depth confidence might be suitable at the cost of highly interpolated points (that could give a wavy look of some surfaces). Note: it is possible to change confidence level in post-processing.</string>
</dict>
<dict>
<key>DefaultValue</key>
@@ -86,7 +86,7 @@
<key>Type</key>
<string>PSGroupSpecifier</string>
<key>FooterText</key>
<string>Smooth the point clouds.</string>
<string>Smooth the point clouds. Note that this can be changed in post-processing.</string>
</dict>
<dict>
<key>Type</key>
@@ -98,6 +98,44 @@
<key>DefaultValue</key>
<false/>
</dict>
<dict>
<key>Type</key>
<string>PSGroupSpecifier</string>
<key>FooterText</key>
<string>Filter interpolated depth pixels between foreground and background objects. Note that values can be changed in post-processing.</string>
</dict>
<dict>
<key>Type</key>
<string>PSMultiValueSpecifier</string>
<key>Title</key>
<string>Depth Bleeding Filter Error</string>
<key>Key</key>
<string>DepthBleedingError</string>
<key>DefaultValue</key>
<string>0</string>
<key>Titles</key>
<array>
<string>Disabled</string>
<string>2.5 cm</string>
<string>5 cm</string>
<string>10 cm</string>
<string>15 cm</string>
<string>20 cm</string>
<string>25 cm</string>
<string>30 cm</string>
</array>
<key>Values</key>
<array>
<string>0</string>
<string>0.025</string>
<string>0.05</string>
<string>0.10</string>
<string>0.15</string>
<string>0.20</string>
<string>0.25</string>
<string>0.30</string>
</array>
</dict>
<dict>
<key>Type</key>
<string>PSGroupSpecifier</string>
@@ -590,7 +628,7 @@
<key>Type</key>
<string>PSGroupSpecifier</string>
<key>FooterText</key>
<string>BRIEF features are fast to compute but are not rotation invariant like FREAK. Warning: Changing feature type will automatically reset the map!</string>
<string>BRIEF, FREAK and ORB are binary features, which are fast to compute and use less memory. SURF and SIFT are high dimensional float descriptors (rotation/scale/shear-invariant), so they use more memory and CPU but they are significantly better to detect loop closures in large environments. Warning: Changing feature type will automatically reset the map!</string>
</dict>
<dict>
<key>Type</key>
@@ -603,13 +641,21 @@
<string>6</string>
<key>Titles</key>
<array>
<string>BRIEF</string>
<string>FREAK</string>
<string>SURF</string>
<string>SIFT</string>
<string>GFTT-FREAK</string>
<string>GFTT-BRIEF</string>
<string>GFTT-ORB</string>
<string>ORB-OCTREE</string>
</array>
<key>Values</key>
<array>
<string>6</string>
<string>0</string>
<string>1</string>
<string>5</string>
<string>6</string>
<string>8</string>
<string>10</string>
</array>
</dict>
<dict>