Darwin
1.10(beta)
|
Encapsulates a single instance of an image for multi-class pixel labeling problems (i.e., image segmentation). More...
Public Member Functions | |
drwnSegImageInstance (const char *imgFilename, const char *baseName=NULL) | |
create a drwnSegImageInstance from file | |
drwnSegImageInstance (const cv::Mat &img, const char *baseName=NULL) | |
create an existing image | |
drwnSegImageInstance (const drwnSegImageInstance &instance) | |
copy constructor | |
virtual | ~drwnSegImageInstance () |
destructor | |
const cv::Mat & | image () const |
returns the colour image | |
const cv::Mat & | greyImage () const |
returns a greyscale version of the image | |
const cv::Mat & | edgeMap () const |
returns the edge magnitude of the image | |
const string & | name () const |
returns the name of the image (if available) | |
int | width () const |
returns the width of the image in pixels | |
int | height () const |
returns the height of the image in pixels | |
int | size () const |
returns the number of pixels in the image | |
int | pixel2Indx (const cv::Point &p) const |
convert from image coordinates to pixel index | |
int | pixel2Indx (int x, int y) const |
convert from image coordinates to pixel index | |
cv::Point | indx2Pixel (int indx) const |
convert from pixel index to image coordinates | |
void | clearPixelFeatures () |
clear cached pixel features or unary terms | |
void | appendPixelFeatures () |
append standard pixel features to cached pixel feature vectors | |
void | appendPixelFeatures (drwnSegImagePixelFeatures &featureGenerator) |
append pixel features to cached pixel feature vectors | |
drwnSegImageInstance & | operator= (const drwnSegImageInstance &instance) |
assignment operator | |
Public Attributes | |
vector< vector< double > > | unaries |
cached unary potentials or pixel feature vectors | |
drwnPixelNeighbourContrasts | contrast |
neighborhood contrast for pairwise smoothness | |
vector< drwnWeightedPixelEdge > | auxEdges |
auxiliary (long range) edges | |
drwnSuperpixelContainer | superpixels |
superpixels for features or consistency terms | |
vector< cv::Mat > | auxiliaryData |
auxiliary data for certain models (future extensions) | |
MatrixXi | pixelLabels |
pixel labels (0 to K-1) and -1 for unknown | |
Protected Member Functions | |
void | initInstance () |
initialization | |
Protected Attributes | |
string | _baseName |
image identifier | |
cv::Mat | _img |
3-channel RGB image of the scene | |
cv::Mat | _grayImg |
1-channel greyscale image of the scene | |
cv::Mat | _softEdgeImg |
1-channel edgemap image of the scene | |
Encapsulates a single instance of an image for multi-class pixel labeling problems (i.e., image segmentation).
The data-structure includes basic image features and can be annotated with pixelwise class labels. Negative labels are assumed to be unknown.