pietrko's profile - activity

Hi, I would like to implement Hough clustering method described in this
paper.

The whole idea is can be summarized in following way.

1. Do typical nearest-neighbour matching between set of features from template and image. Obtaining set SM of matched features.
2. Build 4-dimmensional grid parametrized by object position, scale and rotation, each grid element can contain a feature.
3. Assign features from SM to grid "points" using feature's internal position, orientation etc.
4. Pick the grid point that has the most features in it (represents most probable position).

Any suggestion how could I implement this in fastest and easiest way?

Also can I make sparse matrices in openCV with values being arbitrary classes? Or it has to be numerical type?

Hi, are there any online resources for mathematical side of 360 phtography and imaging? I need to know:

• used projections (360x360 image is not rectangular, what projections are used to store it)?
• problems that arise
• file formats
• math & stuff

Regards, Peter

Hi, I have a following problem that could use community wisdom.

I need to design a real word pattern/image that i can put on the floor or stick to wall, that has following features:

• can easily be detected by opencv, even if highly warped
• I should easily calculate camera matrix from it
• must hold small amount of data (like a number from 0-2^16)
• there will be many such markers in one image (with different data coded in them)

I was thinking QR codes but they are kinda hard to detect, maybe some combination of chessboard and QR inside?

Maybe QR will be ok for that? Which algorithm you suggest then? (Remember it has to deal with multiple qr on the same image and get the camera matrix from it)

FIXED: It was gcc. For some magical, unknown reason gcc 5.2.0 can't link code I've posted properly, whereas with gcc 5.3.x everything is ok. Mark as fixed.

Find the reason it was gcc (can't post answer I'm new -- need to wait 2 days), here is output you asked, highgui is not missing;

Thanks for interest - if you have any idea why gcc 5.3.x is ok, and gcc 5.2.0 gives and error please share.

-lopencv_calib3d -lopencv_contrib -lopencv_core -lopencv_features2d -lopencv_flann -lopencv_gpu -lopencv_highgui -lopencv_imgproc -lopencv_legacy -lopencv_ml -lopencv_nonfree -lopencv_objdetect -lopencv_ocl -lopencv_photo -lopencv_stitching -lopencv_superres -lopencv_ts -lopencv_video -lopencv_videostab -ltbb -lGL -lGLU -lrt -lpthread -lm -ldl

I got linker error trying to compile this simple code:

#include <opencv2/core/core.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <iostream>
#include <string>

using namespace cv;

int main(int argc, char** argv) {

    std::string filename = argv[1];
    Mat A;
    A = imread(filename);
    return 0;
}

Here is the makefile:

CXX=g++
LDFLAGS=`pkg-config --libs opencv`
CXXFLAGS=-Wall -v
CXXFLAGS+=`pkg-config --cflags opencv`


cam_calibration.o : cam_calibration.cxx
    $(CXX) $(CXXFLAGS) $(LDFLAGS) cam_calibration.cxx

clean : 
    rm *.o

The error is following (after issuing "make"):

/tmp/cccbWfsF.o: In function `main':
cam_calibration.cxx:(.text+0x6b): undefined reference to `cv::imread(std::string const&, int)'
collect2: error: ld returned 1 exit status

What is wrong with this code? I'm using Archlinux, g++ (5.2.0) and have all the opencv libraries in version 2.4