1. Background
LSV supports the loading of osgb format oblique photography models generated by DJI Maps and Smart3D (Context Capture), but sometimes the model will be offset when loading a large area model, and the higher the latitude is. The phenomenon becomes more obvious.
It may not be possible to see his specific offset by just looking at it this way. We lock the angle of view to vertical and zoom in to take a closer look at the local situation:
This is the relative center position of the model. We can see that the roads are basically the same according to the satellite image, and there is no obvious deviation. However, when we look at the two ends of the model, the effect is as follows:
It can be clearly seen that there are large deviations. These deviations are not only in the coordinates, but also in the height.
After determining that the terrain height difference is negligible, we can see this phenomenon after moving the model down as a whole: the northern part of the model has almost fallen underground. It can be seen that the model also has some deviation in height.
With this kind of screenshot, we can only roughly see that there is an offset, so what are the specific offset and the reason for the offset?
Second, cause analysis
In order to facilitate further analysis, we must first find out where the specific offset is. Here we use the standard cable data obtained in the skyline as a benchmark, and perform an analysis and comparison after vectorizing the tower in the model.
This is the comparison between our vectorized tower base and the reference cable. In the correct case, the tower base should be between the cables at both ends. Here we take screenshots from west to east (from left to right in the figure):
Through these comparisons, we can verify the conclusions drawn from our previous observations: the offset from the center to the ends is getting larger and larger.
Here we use a text editor to open the position information file metadata.xml of the tilt model:
You can see the projected coordinate system and center point coordinates of the model, and then use a text editor to open the lfp index file generated by LSV:
It can be concluded by calculation that the position field value in the lfp generated in the LSV is the wgs84 latitude and longitude coordinates calculated from the coordinates in the metadata according to the projection coordinate information, and the coordinates are the center coordinates of the model.
Here we can draw the conclusion that because the current LSV only reads the center coordinates of the model, and does not perform the corresponding projection deformation processing on the model, it will cause such an offset, and it will be offset due to the greater the distance from the center of the model. It is also larger, and according to the projection principle of UTM, it can be confirmed that the farther the model is from the equator, the greater the deviation.
Three, the solution
In the current version of LSV, the direct recognition of model projection information is not yet supported, so how can we load a large range of models into the 3D map without offset? Here you can use our LSV Professional Edition. The LSV Professional Edition can directly read the projection information of the model. After loading, it can automatically perform projection deformation processing on the model:
Here we can compare the LSV free version with the LSV professional version's lfp index file (the left side is the free version, the right side is the professional version):
It can be seen here that the professional version loads the projection information of one more model than the free version. It is also because the professional version can automatically recognize the projection information and perform projection deformation processing, so the professional version can achieve no offset compared to the satellite image. , Model loading without deformation.
Here we will conduct a verification by comparing the vectorized power tower with the cable before:
It can be clearly seen here that the bottom surface of all vectorized towers is in the middle of the cable, which is in line with the actual situation. It is concluded that the professional version can better accurately load large-area oblique photography models.