MTF
Public Member Functions | Public Attributes | List of all members
KLDParams Struct Reference
Inheritance diagram for KLDParams:
AMParams ImgParams

Public Member Functions

 KLDParams (const AMParams *am_params, int _n_bins, double _pre_seed, bool _partition_of_unity, bool _debug_mode)
 decides whether logging data will be printed for debugging purposes; More...
 
 KLDParams (const KLDParams *params=nullptr)
 default/copy constructor
 
- Public Member Functions inherited from AMParams
 AMParams (int _resx, int _resy, double _grad_eps=GRAD_EPS, double _hess_eps=HESS_EPS, bool _uchar_input=UCHAR_INPUT, double _likelihood_alpha=AM_LIKELIHOOD_ALPHA, double _likelihood_beta=AM_LIKELIHOOD_BETA, bool _dist_from_likelihood=AM_DIST_FROM_LIKELIHOOD, double _learning_rate=AM_LEARNING_RATE, IlluminationModel *_ilm=nullptr)
 
 AMParams (const AMParams *am_params=nullptr)
 
- Public Member Functions inherited from ImgParams
 ImgParams (int _resx, int _resy, double _grad_eps=GRAD_EPS, double _hess_eps=HESS_EPS, bool _uchar_input=UCHAR_INPUT)
 
 ImgParams (const ImgParams *img_params=nullptr)
 

Public Attributes

int n_bins
 
bool partition_of_unity
 no. More...
 
double pre_seed
 decides whether the partition of unity constraint has to be strictly observed for border bins; More...
 
bool debug_mode
 initial value with which each bin of the joint histogram is pre-seeded More...
 
- Public Attributes inherited from AMParams
double likelihood_alpha
 multiplicative and additive factors for the exponent in the likelihood
 
double likelihood_beta
 
bool dist_from_likelihood
 use negative of likelihood as the distance measure
 
double learning_rate
 optional factor to control the rate of online learning
 
ILM ilm
 optional parametric function of pixel values that can account for lighting changes
 
- Public Attributes inherited from ImgParams
int resx
 horizontal and vertical sampling resolutions
 
int resy
 
double grad_eps
 numerical increment/decrement used for computing image hessian and gradient using the method of finite differences
 
double hess_eps
 
bool uchar_input
 

Additional Inherited Members

- Public Types inherited from AMParams
typedef shared_ptr< IlluminationModelILM
 

Constructor & Destructor Documentation

KLDParams::KLDParams ( const AMParams am_params,
int  _n_bins,
double  _pre_seed,
bool  _partition_of_unity,
bool  _debug_mode 
)

decides whether logging data will be printed for debugging purposes;

only matters if logging is enabled at compile time value constructor

Member Data Documentation

bool KLDParams::debug_mode

initial value with which each bin of the joint histogram is pre-seeded

to avoid numerical instabilities due to empty or near empty bins (caused e.g. by having to assume log(0) = 0 for empty bins)

bool KLDParams::partition_of_unity

no.

of bins in the histograms used internally - dimensionality of the KLD error vector will be n_bins * n_bins; if partition_of_unity is enabled, this should be 2 more than the desired no. of bins (w.r.t normalized pixel range) since the actual range within which the pixel values are normalized is 2 less than this value to avoid boundary conditions while computing the contribution of each pixel to different bins by ensuring that pixels with the maximum and minimum values contribute to all 4 bins required by the bspl function of degree 3 used here;

double KLDParams::pre_seed

decides whether the partition of unity constraint has to be strictly observed for border bins;

if enabled, the pixel values will be normalized in the range [1, n_bins-2] so each pixel contributes to all 4 bins


The documentation for this struct was generated from the following file: