-
Notifications
You must be signed in to change notification settings - Fork 15
/
num_intrinsics_optimization_params.docstring
63 lines (45 loc) · 2.47 KB
/
num_intrinsics_optimization_params.docstring
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
Get the number of optimization parameters for a single camera's intrinsics
SYNOPSIS
m = mrcal.cameramodel('xxx.cameramodel')
f( m.optimization_inputs() )
...
def f(optimization_inputs):
Nstates = mrcal.num_intrinsics_optimization_params(**optimization_inputs)
...
Return the number of parameters used in the optimization of the intrinsics of a
camera.
The optimization algorithm sees its world described in one, big vector of state.
The optimizer doesn't know or care about the meaning of each element of this
vector, but for later analysis, it is useful to know what's what.
This function reports how many optimization parameters are used to represent the
intrinsics of a single camera. This is very similar to
mrcal.lensmodel_num_params(), except THIS function takes into account the
do_optimize_intrinsics_... variables used to lock down some parts of the
intrinsics vector. Similarly, we have mrcal.num_states_intrinsics(), which takes
into account the optimization details also, but reports the number of variables
needed to describe ALL the cameras instead of just one.
In order to determine the variable mapping, we need quite a bit of context. If
we have the full set of inputs to the optimization function, we can pass in
those (as shown in the example above). Or we can pass the individual arguments
that are needed (see ARGUMENTS section for the full list). If the optimization
inputs and explicitly-given arguments conflict about the size of some array, the
explicit arguments take precedence. If any array size is not specified, it is
assumed to be 0. Thus most arguments are optional.
ARGUMENTS
- **kwargs: if the optimization inputs are available, they can be passed-in as
kwargs. These inputs contain everything this function needs to operate. If we
don't have these, then the rest of the variables will need to be given
- lensmodel: string specifying the lensmodel we're using (this is always
'LENSMODEL_...'). The full list of valid models is returned by
mrcal.supported_lensmodels(). This is required if we're not passing in the
optimization inputs
- do_optimize_intrinsics_core
do_optimize_intrinsics_distortions
do_optimize_extrinsics
do_optimize_calobject_warp
do_optimize_frames
optional booleans; default to True. These specify what we're optimizing. See
the documentation for mrcal.optimize() for details
RETURNED VALUE
The integer reporting the number of optimization parameters used to describe the
intrinsics of a single camera