Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

Error with OpenCV ROI

Tags:

c++

opencv

I am trying to make a program that will identify a pattern by sliding a ROI over an image and comparing the ROI with a template, it will compare pixel values of the ROI and the template and increase a counter by 1 each time a pixel matches, then I compare the counter with a threshold, if it passes a rectangle will be drawn, if not, it will continue to slide trough the image, if I run the debugger on it, it shows no errors sliding trough the image, but if i run it normally it throws the next exception:

OpenCV Error: Assertion failed (0 <= roi.x && 0 <= roi.width && roi.x + roi.width <= m.cols && 0 <= roi.y && 0 <= roi.height && roi.y + roi.height <= m.rows) in Mat, file /home/abuild/rpmbuild/BUILD/opencv-2.4.6.1/modules/core/src/matrix.cpp, line 323
terminate called after throwing an instance of 'cv::Exception'
what():  /home/abuild/rpmbuild/BUILD/opencv-2.4.6.1/modules/core/src/matrix.cpp:323: error: (-215) 0 <= roi.x && 0 <= roi.width && roi.x + roi.width <= m.cols && 0 <= roi.y && 0 <= roi.height && roi.y + roi.height <= m.rows in function Mat

I leave the code Bellow :

#include <stdio.h>
#include <stdlib.h>
#include <iostream>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/core/core.hpp>
#include <opencv2/imgproc/imgproc.hpp>


using namespace std;
using namespace cv;

Mat         iOrig;                          //imagen que se analizara
Mat         patron;                         //patron buscado
Point       corner1,corner2;                //esquinas del ROI
bool        tresholdpass;                   //boolean returned by comparar()
Rect        box;                            //rectangulo usado para crear el ROI
float       porcentaje;                     //distancia a recorrer que sera dada por el tamano de la imagen muestra para el patron
float       treshold;                       //treshold que debera ser superado para que se considere similar o igual al patron
float       valpx;                          //valor almacenado en el pixel comparado
float       contTreshold;                   //contador a comparar contra el porcentaje requerido para considerarse como patron encontrado
float       totpxpat;                       //cantidad de pixeles de la imagen muestra del patron
float       porctreshold;                   //porcentaje representativo para considerar que se encontro el patron en esa ROI

bool comparar(Mat region, Mat patron){

    int i=0;
    int j=0;
    contTreshold=0;

    for (i=0;i<patron.cols;i++){
        for (j=0;j<patron.rows;j++){
            Point a(i,j);
            if(abs(region.at<float>(a))==abs(patron.at<float>(a))){
            //se compara el contenido de el ROI y el patron en un punto
                contTreshold++;             //en caso de ser cierto, el contador aumenta en 1
            }
        }

    }
    totpxpat = patron.rows*patron.cols;     //se cuentan la cantidad de pixeles dado columnas*renglones
    porctreshold = 0.8;                     //el porcentaje que se usara para el treshold
    treshold = totpxpat * porctreshold;     //el treshold que determinara si se cumple el porcentaje de la imagen
                                        //para saber si cumple el patron o no, en caso de q se supere, retornara verdadero
                                        //en caso de que no se supere retornara falso y se elegira otro ROI para analizar
    if (contTreshold>treshold){
        return true;
    }else{
        return false;
    }
}

int main() {

    namedWindow("imagen");
    namedWindow("segmento");

    iOrig = imread( "/home/diego/Downloads/figuras.jpg",CV_LOAD_IMAGE_GRAYSCALE );
    patron = imread("/home/diego/Downloads/patron.jpg",CV_LOAD_IMAGE_GRAYSCALE);

    imshow("imagen",iOrig);
    imshow("segmento",patron);
    corner1.x = 1;
    corner1.y = 1;
    corner2.x = patron.cols;
    corner2.y = patron.rows;


    porcentaje = (int)patron.cols * 0.05;

    while (corner2.x<iOrig.rows-(patron.rows*2)){
        while(corner2.y<iOrig.cols-(patron.cols*2)){
            box.width = abs (corner1.x-corner2.x)+1;
            box.height = abs (corner1.y - corner2.y)+1;
            box.x = min(corner1.x, corner2.x);
            box.y = min(corner1.y, corner2.y);


        //se crea una imagen de la region de interes seleccionada apartir de las 2 esquinas de la ROI
            Mat     region(iOrig,box);              //region de interes que sera comparada

        //se manda a comparar el ROI con el patron
            tresholdpass=comparar(region,patron);
            if (tresholdpass == true){
                Mat local_img = iOrig.clone();
                rectangle(local_img,corner1,corner2,Scalar(0,0,255));
                imshow("imagen",local_img);

            }
            corner1.x+=porcentaje;
            corner2.x+=porcentaje;
        }
        corner1.y+=porcentaje;
        corner2.y+=porcentaje;
    }

    while (char(waitKey(1))!= 'q'){}
    return 0;
}

I cant upload the images im using because of the reputation... but the original image is 800 x 450 and the template searched in the image is 131 x 132

sorry about the comments in my code being in Spanish, English is not my native language as you have guessed by now, i really don't know where my mistake is but... I hope it is simple thank you in advance!

like image 845
Diego Hernandez Reyes Avatar asked Apr 18 '15 01:04

Diego Hernandez Reyes


1 Answers

It means you're trying to get the ROI region out of the image plane. You should make sure the ROI area to be within the image plane in order to avoid the crash.

For your case, you can do like this:

// check the box within the image plane
if (0 <= box.x
    && 0 <= box.width
    && box.x + box.width <= iOrig.cols
    && 0 <= box.y
    && 0 <= box.height
    && box.y + box.height <= iOrig.rows){
    // box within the image plane
    Mat region(iOrig, box);
}
else{
    // box out of image plane, do something...
}
like image 131
herohuyongtao Avatar answered Sep 20 '22 02:09

herohuyongtao