|
Darwin
1.10(beta)
|
All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Groups Pages
|
Darwin
1.10(beta)
|
Utility class for computing multiple template matches. More...
Public Member Functions | |
| drwnTemplateMatcher () | |
| default constructor | |
| drwnTemplateMatcher (const drwnTemplateMatcher &tm) | |
| copy constructor | |
| void | clear () |
| clears all templates and buffered storage | |
| bool | empty () const |
| returns true if there are no templates | |
| int | size () const |
| returns the number of templates | |
| int | width () const |
| returns the width of the largest template | |
| int | height () const |
| returns the height of the largest template | |
| void | reset () |
| release internal memory (keeps templates) | |
| void | addTemplate (cv::Mat &t) |
| Add templates to the object. The difference between addTemplate and copyTemplate is that the former takes does not make a copy so changes by the calling function will be corrupt the data. Templates should be single channel 32-bit floating point. | |
| void | addTemplates (vector< cv::Mat > &t) |
| See above. | |
| void | copyTemplate (const cv::Mat &t) |
| See above. | |
| void | copyTemplates (const vector< cv::Mat > &t) |
| See above. | |
| vector< cv::Mat > | responses (const cv::Mat &img, int method=CV_TM_CCORR) |
| Response images are all generate to be the same size as the original image. Boundaries are zero-padded. Caller is responsible for freeing memory. Functions without image argument use the previously loaded image. | |
| vector< cv::Mat > | responses (int method=CV_TM_CCORR) |
| See above. | |
| cv::Mat | response (const cv::Mat &img, unsigned tid, int method=CV_TM_CCORR) |
| See above. | |
| cv::Mat | response (unsigned tid, int method=CV_TM_CCORR) |
| See above. | |
| drwnTemplateMatcher & | operator= (const drwnTemplateMatcher &tm) |
| assignment operator | |
| const cv::Mat & | operator[] (unsigned i) const |
| returns the i-th template | |
Protected Member Functions | |
| void | cacheDFTAndIntegrals (const cv::Mat &image) |
| caches all buffered data for a given image | |
Protected Attributes | |
| vector< cv::Mat > | _templates |
| vector of templates to match | |
| vector< pair< double, double > > | _templateStats |
| template mean and energy | |
| cv::Size | _largestTemplate |
| dimensions of the largest template | |
| cv::Size | _imgSize |
| size of image being buffered | |
| vector< cv::Mat > | _dftTemplates |
| buffered templates in fourier space | |
| cv::Mat | _dftImage |
| buffered image in fourier space | |
| cv::Mat | _dftResponse |
| buffer for response image | |
| cv::Mat | _imgSum |
| buffer for integral image | |
| cv::Mat | _imgSqSum |
| buffer for integral square image | |
Utility class for computing multiple template matches.
The instantiated object pre-computes template (and image) DFTs to allow for faster matching on multiple images (of the same size). However, the code is very memory intensive. Call the drwnTemplateMatcher::reset function to release the internal buffers.
1.8.6