ggeo's profile - activity

@pklab:Ok, thanks.

@pklab:You meant int instead of 1 above?So,even if I declare test as double type ,it will be uchar type.

@pklab:Ok,thanks for the help!

@berak: Hmm, so even though I am declaring test matrix to be double , the comparison results in uchar.But the type of test matrix remains double?Thanks! ( make that an answer please)

I want to know why trying to access (and print) test matrix results in showing garbage:

Mat A = (Mat_<double>(3,3) << 0, -1.4, 0.2, -1, 5.4, -1, 0.1, -1.4, 0);
cout << "A = " << endl << " " << A << endl << endl;

Mat B = (Mat_<double>(3,3) << 0, -1.4, 0.4, -2, 5.4, -11, 0, -1.42, 0);
cout << "B = " << endl << " " << B << endl << endl;

Mat test = Mat(3,3,CV_64F);

test = ( A == B ); 

for ( int i = 0; i < 3; i++ )     
{         
    for (int j = 0; j < 2; j++ )         
    {             
        cout << test.at<int>(i,j) << endl;
    }             
}

If I use just cout << test it is ok ,but I want to know why the previous doesn't work.

Thanks!

(This is related to this )

I have a vector<Mat> src which is filled with data. It's size is src.resize( N )

I have another vector<mat> dest; I am declaring it's size to be dest.resize( N )

1) What is the right way to copy src to dest?Just dest=src?

2) If I want to copy to dest ,only the 1st row and all columns of src vector, then what size the dest vector should be?

dest = src;
src[0]( Range(0, 1),Range::all() ).copyTo( dest[0](Range::all(), Range::all() ) ); //for the 1st array of vector

If ,for example src is 4x4 , then ,since I want to keep only 1st row and all columns,the dest size should be 1x4. Is there a way to do this ?Using the src[0]( Range(0, 1),Range::all() ).copyTo( dest[0](Range::all(), Range::all() ) )?

Or,another idea , is to have the dest the same size with src ,but fill all the rest elements with zero for example?

=====UPDATE================

I figured this:

    dest[0] = src[0 ]( Range(0, 1),Range::all() );

for first row and all columns.

( I can't make it an answer yet ,but if there is another way please share )

The truth is that I have vector<Mat> src and vector<Mat> dst;Both , with .resize(N).I tried copy dst[0]= src[0];but it gives me floating point exception. ( src is a filled vector and dst empty).Thanks!

Thanks for the help.I am trying the last methid you described but I am receiving assertion failed...m.dims >= 2 in function Mat. My src mat has the same size as dst.But , keep in mind that I want to copy(keep) only a row and all columns from src to the dest.So, what about the size of dest?Just be the same as src?

I can't figure how to copy a part of mat to another mat. ( or to itself)

I want for example to copy (or to keep) only the first row of srcMat.

I am using :

    for ( int x = 0; x < height; x++ )
    {
        for ( int y = 0; y < width; y++ )
        {
            destMat = scrMat.at<float>(0,y);
           // or destMat.at<float>(x,y) = scrMat.at<float>(0,y);
        }
    }

I am not sure about using:

            srcMat.copyTo( destMat( Rect( y, 0, srcMat.cols, srcMat.rows ) ) );

@thdrksdfthmn:Yes,that is ok, thanks!

@LorennaGdL:Ok, thanks!( I was a little confused)

@LorenaGdL:But , I am setting the mycompare[0] , access it in the countNonZero function ,then at the 2nd loop , set the mycompare[1] , access it and so on. ( I am just asking for the if statement )

@LorenaGdL:If I just do mycompare[ index++ ] = ( A[ idx ] == A[ k ] ); myCounts[ index ] = countNonZero( mycompare[ index ] ); , should't work also ? ( because it doesn't)

@LorenaGdL:Yes!You are right,I missed that!Is there a way to use if statement for this? ( please make it an answer if you want)Thanks!

Hello , I want to use countNonZero , but I am getting a seg fault at the countNnZero line.

vector<Mat> mycompare;
int compareSize = (N / 2) * (N -1);
mycompare.resize( compareSize );

int myCounts [ compareSize ]; 

int index = 0;
for ( int idx = 0; idx < N; idx++ )
{
    for ( int k = idx + 1; k < N; k ++ )
    {
         mycompare[ index++ ] = ( A[ idx ] == A[ k ] );
         myCounts[ index++ ] = countNonZero( mycompare[ index ] );
     }
}

OR

Is there a way to use if statement?:

if ( mycompare[ index ] == 255 ) 
    count++;

@LorenaGdL:Ok.So , mat objects do element by element operations.Ok.Thanks.If you want make it an answer.

@LorenaGdL:But still I am going to access .at<float>( coordinates) ,right?Can you give an example?

@LorenaGdL:I just want to know why this happens. You said : f you're gonna use a per-pixel loop later, avoid it....And then ,what should I use?

@LorenaGdL:That's right.But , why can I access at<float>( 300,239999 ) ??I would expect until at<float>( 199,199).Thanks

@LorenaGdL:Hello ,I updated.I am scanning in the loop the whole image (600x600) and I am creating 9 pieces of 200x200. ( 800/200 = 3)

@sturkmen:Thank's but I am trying to understand what is wrong with my code.

I have a 600x600 image ( Image) and i am cutting into pieces of 200x200 (piecesImages),by creating 3x3 blocks (Height = 3 , Width =3).

Mat pieceImageFrame (200,200,CV_32F);
vector<Rect> piecesImagesRois;
vector<Mat> piecesImages;
piecesImagesRois.reserve( Width * Height );
piecesImages.reserve( Width * Height );

for( int i = 0; i <ImageSize.height; i+= pieceImageSize.height ) //ImageSize.height = 600
{
    for( int j = 0; j < ImageSize.width; j+= pieceImageSize.width )//ImageSize.width = 600
    {
        Rect myrect = Rect( j,i, pieceImageSize.width, pieceImageSize.height );
        pieceImageFrame = Image( myrect );
        piecesImages.push_back( pieceImageFrame.clone() );
        piecesImagesRois.push_back(myrect);             
    }
}

Now, I would expect to be able to access every image until :

piecesImages[ 0 ].at<float>( 199,199)

but as I can see I am able to access:

piecesImages[ 0 ].at<float>( 300,239999 )  !!

I would expect a grabage value but the values seem ok!

@pklab:Thanks for the help.So,the proper way to use pointers in order to access data is (after checking the continuous ) is to use const float* ptr1 = grayImage.ptr<float>(i);in the loop ?And what about accessing the data outside the loop? ( because ptr1 is defined in the loop according to [this] http://docs.opencv.org/3.0.0/d3/d63/c...

Hello , I just started to study openCV and I am trying to figure how to properly read the image as array.

First of all , when I am loading the image using imread , what is the type ? Always unsigned char?

So,for example, I am trying to do:

...
Mat  grayImage;
...

unsigned char *myData = grayImage.data;
int width = grayImage.cols;
int height = grayImage.rows;

for(int i = 0; i < height; i++)
{
    for(int j = 0; j < width; j++)
    {
        cout << "Mat[ " << i * width + j << " ] = " << (myData[ i * width + j ]) << "\t";
    }
    cout << endl;
}

I am receiving as output symbols and letters.

Also, if I want to store an array as an image what steps must I follow?

(I forgot to mention that I want to use pointers in order to access data)