Running Sift feature extraction in VS environment (principle part)

/*

If you give two pictures, there are similarities in the middle. Required to make a match. What to do. Can I tell you now? For example, give two pictures, first find out the sift point. Scale space extremum detection. The key point orientation of Gaussian fuzzy key point locating key points description Kdtree and BBF optimal node priority algorithm to match two picture feature points, will cover some incorrect matching points RANSAC random sampling consistent, eliminate inappropriate points to match the point, limit to a certain right place Based on the results of this match. Determines the point at which the two images intersect. For example, the overlap of the two images is, the upper-left and bottom-right way, then in the non-overlapping place, according to the left image to write, and then All Black. The overlapping places, the smooth transition according to the weight value. The lower left part is blacked out and the bottom right corner does not overlap in the right image to write.

*/

Take it easy:

Total of six files in three groups

Imgfeatures.h and IMGFEATURES.C Parts

Enumeration Type 1:

Feature_type

Enumeration Type 2:

Feature_match_type

Two sets of feature colors

#define Feature_oxfd_color Cv_rgb (255,255,0)
#define Feature_lowe_color Cv_rgb (255,0,255)

Length of the sub-paragraph

#define FEATURE_MAX_D 128

Feature structure

Feature

Four functions:

1. Import feature points

2. Export feature points

3. Feature points of painting

4. Calculate the Euclidean distance between the two sub-references.

When describing a feature point import: http://www.robots.ox.ac.uk:5000/~vgg/research/affine/index.html

There is such a note

Then the implementation

When importing feature points:

Reading feature points from a file, the file should be organized in the form of an Oxford visual geometry, or in accordance with the Davirol encoding format. TYPE The parameter of this place if it is feature_type then she will be treated according to the Oxford characteristic input file, that is, the Oxford visual geometry, see http://www.robots.ox.ac.uk:5000/~vgg/research/ Affine/index.html

Breakpoint: Affine covariant feature

Overview: This page is aimed at detecting affine invariant features in any image and evaluating the performance of an area detection sub. "Should be the description of how to describe the affine invariant characteristics, as well as the evaluation of the quality of the sub-performance."

Public:

Area detection:

Http://www.robots.ox.ac.uk:5000/~vgg/research/affine/det_eval_files/mikolajczyk_ijcv2004.pdf "Scale affine interest point invariant detection"

Http://www.robots.ox.ac.uk:5000/~vgg/research/affine/det_eval_files/matas_bmvc2002.pdf "Most stable extreme area"

Http://www.robots.ox.ac.uk:5000/~vgg/research/affine/det_eval_files/tuytelaars_ijcv2004.pdf "based on affine invariant region matching wide area segmentation scene"

Http://www.robots.ox.ac.uk:5000/~vgg/research/affine/det_eval_files/kadir04.pdf "Radiation invariant significant area detection"

All Detectors-survey: T. tuytelaars and K. Mikolajczyk , Local invariant Feature detectors- Survey. In CVG, 3 (1): 1-110, 2008. "The scientific name is this, if found, please contact ~ ~ ~"

Then this fifth document can not be found, I hope that the high play together to help find the next

Area Description:

Http://www.robots.ox.ac.uk:5000/~vgg/research/affine/det_eval_files/lowe_ijcv2004.pdf "unique image features from scale invariant key points"

Performance evaluation:

Http://www.robots.ox.ac.uk:5000/~vgg/research/affine/det_eval_files/vibes_ijcv2004.pdf "The contrast of affine area detectors"

Http://www.robots.ox.ac.uk:5000/~vgg/research/affine/det_eval_files/mikolajczyk_pami2004.pdf "Local Sub performance evaluation"

Software:

There is no Linux, and no matlab do not want to get ...

Test data:

You can download it.

IMGFEATURES.C is the process of implementation.

Then look at the sift.h the inside of the thing is also very concise, suddenly think these things a bit like Java inside the public function: C inside the inverted are more like the private function can only be used inside it.

Look what's in the sift.h.

Declaration of a large pile of variables and two functions

extern int sift_features (iplimage* img, struct feature** feat);

extern int _sift_features (iplimage* img, struct feature** feat, int intvls,
Double Sigma, double contr_thr, int curv_thr,
int img_dbl, int descr_width, int descr_hist_bins);

and seems to be in the back or call relationship, after all, two names are very similar. And then a little bit of the discovery, is called the same function.

So sift's two files are used to find the SIFT feature points. But how to find it. The next step may be to look at the contents of this function.

Running Sift feature extraction in VS environment (principle part)