I have the following code:
//newImg is a mat of an image and orderedKeyPoint is the result from Fast
cv::FREAK extractor;
cv::Mat queryDescriptors;
extractor.compute(newImg, orderedKeyPoint, queryDescriptors);
I am trying to access individual freak descriptors using queryDescriptors.at< ???>(r,0) where r is an arbitrary valid row value but I am not sure of the type. All documentation states that it is just a descriptor, but is that of type Mat or double or something else? Is this the best way of doing it?
cv::Mat descriptor2 = queryDescriptors.at<cv::Mat>(2,0);
I would like to be able to reconstruct queryDescriptors (Mat of descriptors) from individual descriptors by taking them and putting them in the row values of a cv::Mat, ex:
queryDescriptors.at<cv::Mat>(2,0) = descriptor2;
Any help or insight would be greatly appreciated,
Isaac
the FREAK descriptor is a uchar Mat with 64 cols and numkeypoints rows.
so, to get to an element of it:
uchar elm = descriptor.at<uchar>(row,col);
where row is the keypoint id, and col is the element id.
If you have a look into \opencv\modules\features2d\src\freak.cpp you can see:
int FREAK::descriptorSize() const
{
return FREAK_NB_PAIRS / 8; // descriptor length in bytes
}
int FREAK::descriptorType() const
{
return CV_8U;
}
int FREAK::defaultNorm() const
{
return NORM_HAMMING;
}
} // END NAMESPACE CV
So uchar seems to be the type as berak already suggested.
Related
What I want is extracting a value from vector using a variable scalar index.
Like _mm_extract_epi8 / _mm256_extract_epi8 but with non-immediate input.
(There are some results in the vector, the one with the given index is found out to be the true result, the rest are discarded)
Especially, if index is in a GPR, the easiest way is probably to store val to memory and then movzx it into another GPR. Sample implementation using C:
uint8_t extract_epu8var(__m256i val, int index) {
union {
__m256i m256;
uint8_t array[32];
} tmp;
tmp.m256 = val;
return tmp.array[index];
}
Godbolt translation (note that a lot of overhead happens for stack alignment -- if you don't have an aligned temporary storage area, you could just vmovdqu instead of vmovdqa): https://godbolt.org/z/Gj6Eadq9r
So far the best option seem to be using _mm_shuffle_epi8 for SSE
uint8_t extract_epu8var(__m128i val, int index) {
return (uint8_t)_mm_cvtsi128_si32(
_mm_shuffle_epi8(val, _mm_cvtsi32_si128(index)));
}
Unfortunately this does not scale well for AVX. vpshufb does not shuffle across lanes. There is a cross lane shuffle _mm256_permutevar8x32_epi32, but the resulting stuff seem to be complicated:
uint8_t extract_epu8var(__m256i val, int index) {
int index_low = index & 0x3;
int index_high = (index >> 2);
return (uint8_t)(_mm256_cvtsi256_si32(_mm256_permutevar8x32_epi32(
val, _mm256_zextsi128_si256(_mm_cvtsi32_si128(index_high))))
>> (index_low << 3));
}
I am aware there are several ways to read and write a pixel value of an OpenCV cv::Mat image/matrix.
A common one is the .at<typename T>(int, int) method http://opencv.itseez.com/2.4/modules/core/doc/basic_structures.html#mat-at .
However, this requires the typename to be known, for instance .at<double>.
The same thing applies to more direct pointer access OpenCV get pixel channel value from Mat image .
How can I read a pixel value without knowing its type? For instance, it would be ok to receive a more generic CvScalar value in return. Efficiency is not an issue, as I would like to read rather small matrices.
Kind of. You can construct cv::Mat_ and provide explicit type for elements, after that you don't have to write element type each time. Quoting opencv2/core/mat.hpp
While Mat is sufficient in most cases, Mat_ can be more convenient if you use a lot of element
access operations and if you know matrix type at the compilation time. Note that
Mat::at(int y,int x) and Mat_::operator()(int y,int x) do absolutely the same
and run at the same speed, but the latter is certainly shorter.
Mat_ and Mat are very similar. Again quote from mat.hpp:
The class Mat_<_Tp> is a thin template wrapper on top of the Mat class. It does not have any
extra data fields. Nor this class nor Mat has any virtual methods. Thus, references or pointers to
these two classes can be freely but carefully converted one to another.
You can use it like this
Mat_<Vec3b> dummy(3,3);
dummy(1, 2)[0] = 10;
dummy(1, 2)[1] = 20;
dummy(1, 2)[2] = 30;
cout << dummy(1, 2) << endl;
Why I said 'kind of' in the first place? Because if you want to pass this Mat_ somewhere - you have to specify it's type. Like this:
void test(Mat_<Vec3b>& arr) {
arr(1, 2)[0] = 10;
arr(1, 2)[1] = 20;
arr(1, 2)[2] = 30;
cout << arr(1, 2) << endl;
}
...
Mat_<Vec3b> dummy(3,3);
test(dummy);
Technically, you are not specifying your type during a pixel read, but actually you still have to know it and cast the Mat to the appropriate type beforehand.
I guess you can find a way around this using some low-level hacks (for example make a method that reads Mat's type, calculates element size and stride, and then accesses raw data using pointer arithmetic and casting...). But I don't know any 'clean' way to do this using OpenCV's functionality.
If you already know the type, you can use Mat_<> type for easy access. If you don't know the type, you can:
convert the data to double, so data won't be truncated in any case
switch over the number of channels to access correctly the double matrix. Note that you can have at most of 4 channels, since Scalar has at most 4 elements.
The following code will convert only the selected element of the source matrix to a double value (with N channels).
You get a Scalar containing the value at position row, col in the source matrix.
#include <opencv2/opencv.hpp>
#include <iostream>
using namespace std;
using namespace cv;
Scalar valueAt(const Mat& src, int row, int col)
{
Mat dst;;
src(Rect(col, row, 1, 1)).convertTo(dst, CV_64F);
switch (dst.channels())
{
case 1: return dst.at<double>(0);
case 2: return dst.at<Vec2d>(0);
case 3: return dst.at<Vec3d>(0);
case 4: return dst.at<Vec4d>(0);
}
return Scalar();
}
int main()
{
Mat m(3, 3, CV_32FC3); // You can use any type here
randu(m, Scalar(0, 0, 0, 0), Scalar(256, 256, 256, 256));
Scalar val = valueAt(m, 1, 2);
cout << val << endl;
return 0;
}
i have a code like this:
Mat img = Highgui.imread(inFile);
Mat templ = Highgui.imread(templateFile);
int result_cols = img.cols() - templ.cols() + 1;
int result_rows = img.rows() - templ.rows() + 1;
Mat result = new Mat(result_rows, result_cols, CvType.CV_32FC1);
Imgproc.matchTemplate(img, templ, result, Imgproc.TM_CCOEFF);
/////Core.normalize(result, result, 0, 1, Core.NORM_MINMAX, -1, new Mat());
for (int i = 0; i < result_rows; i++)
for (int j = 0; j < result_cols; j++)
if(result.get(i, j)[0]>?)
//match!
I need to parse the input image to find multiple occurrencies of the template image. I want to have a result like this:
result[0][0]= 15%
result[0][1]= 17%
result[x][y]= 47%
If i use TM_COEFF all results are [-xxxxxxxx.xxx,+xxxxxxxx.xxx]
If i use TM_SQDIFF all results are xxxxxxxx.xxx
If i use TM_CCORR all results are xxxxxxxx.xxx
How can i detect a match or a mismatch? What is the right condition into the if?
If i normalized the matrix the application set a value to 1 and i can't detect if the template isn't stored into the image (all mismatch).
Thanks in advance
You can append "_NORMED" to the method names (For instance: CV_TM_COEFF_NORMED in C++; could be slightly different in Java) to get a sensible value for your purpose.
By 'sensible', I mean that you will get values in the range of 0 to 1 which can be multiplied by 100 for your purpose.
Note: For CV_TM_SQDIFF_NORMED, it will be in the range -1 to 0, and you will have to subtract the value from 1 in order to make sense of it, because the lowest value if used in this method.
Tip: you can use the java equivalent of minMaxLoc() in order to get the minimum and maximum values. It's very useful when used in conjunction with matchtemplate.
I believe 'minMaxLoc' that is located inside the class Core.
Here's a C++ implementation:
matchTemplate( input_mat, template_mat, result_mat, method_NORMED );
double minVal, maxVal;
double percentage;
Point minLoc; Point maxLoc;
minMaxLoc( result, &minVal, &maxVal, &minLoc, &maxLoc, Mat() );
if( method_NORMED == CV_TM_SQDIFF_NORMED )
{
percentage=1-minVal;
}
else
{
percentage=maxVal;
}
Useful C++ docs:
Match template description along with available methods: http://docs.opencv.org/modules/imgproc/doc/object_detection.html
MinMaxLoc documentation:
http://docs.opencv.org/modules/core/doc/operations_on_arrays.html?highlight=minmaxloc#minmaxloc
Another approach will be background differencing. You can observe the distortion.
import org.opencv.core.Core;
import org.opencv.core.Mat;
import org.opencv.highgui.Highgui;
import org.opencv.imgproc.Imgproc;
public class BackgroundDifference {
public static void main(String[] arg){
System.loadLibrary(Core.NATIVE_LIBRARY_NAME);
Mat model = Highgui.imread("e:\\answers\\template.jpg",Highgui.CV_LOAD_IMAGE_GRAYSCALE);
Mat scene = Highgui.imread("e:\\answers\\front7.jpg",Highgui.CV_LOAD_IMAGE_GRAYSCALE);
Mat diff = new Mat();
Core.absdiff(model,scene,diff);
Imgproc.threshold(diff,diff,15,255,Imgproc.THRESH_BINARY);
int distortion = Core.countNonZero(diff);
System.out.println("distortion:"+distortion);
Highgui.imwrite("e:\\answers\\diff.jpg",diff);
}
}
I am trying to print cv::Mat which contains my image. However whenever I print the Mat using cout, a 2D array printed into my text file. I want to print one one pixel in one line only. How can i print line wise pixels from cv::Mat.
A generic for_each loop, you could use it to print your data
/**
*#brief implement details of for_each_channel, user should not use this function
*/
template<typename T, typename UnaryFunc>
UnaryFunc for_each_channel_impl(cv::Mat &input, int channel, UnaryFunc func)
{
int const rows = input.rows;
int const cols = input.cols;
int const channels = input.channels();
for(int row = 0; row != rows; ++row){
auto *input_ptr = input.ptr<T>(row) + channel;
for(int col = 0; col != cols; ++col){
func(*input_ptr);
input_ptr += channels;
}
}
return func;
}
use it like
for_each_channel_impl<uchar>(input, 0, [](uchar a){ std::cout<<(size_t)a<<", "; });
you could do some optimization to continuous channel, then it may looks like
/**
*#brief apply stl like for_each algorithm on a channel
*
* #param
* T : the type of the channel(ex, uchar, float, double and so on)
* #param
* channel : the channel need to apply for_each algorithm
* #param
* func : Unary function that accepts an element in the range as argument
*
*#return :
* return func
*/
template<typename T, typename UnaryFunc>
inline UnaryFunc for_each_channel(cv::Mat &input, int channel, UnaryFunc func)
{
if(input.channels() == 1 && input.isContinuous()){
return for_each_continuous_channels<T>(input, func);
}else{
return for_each_channel_impl<T>(input, channel, func);
}
}
This kind of generic loopsave me a lot of times, I hope you find it helpful.If there are
any bugs, or you have better idea, please tell me.
I would like to design some generic algorithms for opencl too, sadly it do not support
template, I hope one day CUDA will become an open standard, or opencl will support template.
This works for any number of channels as long as the channels type are base on byte, non-byte
channel may not work.
I am passing a Mat to another function and changing it inside the called function. I had expected that being a more complex type it was automatically passed by reference so that the matrix would have changed in the calling function, but it doesn't. Could someone point me at the explanation of how to correctly return a changed Mat from a function?
Here's the code snippet:
void callingFunction(Mat img)
{
Mat tst(100,500,CV_8UC3, Scalar(0,255,0));
saveImg(tst, "Original image", true);
testImg(tst);
saveImg(tst, "Want it to be same as inside testImg but is same as Original", true);
}
void testImg(Mat img)
{
int rs = 50; // rows
int cs = 100; // columns
img = Mat(rs, cs, CV_8UC3, Scalar(255,0,0));
Mat roi(img, Rect(0, 0, cs, rs/2));
roi = Scalar(0,0,255); // change a subsection to a different color
saveImg(img, "inside testImg", true);
}
Thanks!
You have to define Mat as parameter-reference (&). Here's edited code:
void callingFunction(Mat& img)
{
Mat tst(100,500,CV_8UC3, Scalar(0,255,0));
saveImg(tst, "Original image", true);
testImg(tst);
saveImg(tst, "Want it to be same as inside testImg but is same as Original", true);
}
void testImg(Mat& img)
{
int rs = 50; // rows
int cs = 100; // columns
img = Mat(rs, cs, CV_8UC3, Scalar(255,0,0));
Mat roi(img, Rect(0, 0, cs, rs/2));
roi = Scalar(0,0,255); // change a subsection to a different color
saveImg(img, "inside testImg", true);
}
I wondered about the same question myself, so I would like to further clarify the answer given by #ArtemStorozhuk (which is correct).
The OpenCV documentation is misleading here, because it appears you're passing the matrix by value, but in fact the constructor of cv::OutputArray is defined as follows:
_OutputArray::_OutputArray(Mat& m)
so it gets the matrix by reference!
Since operations like cv::Mat::create create a new matrix, the operation releases the reference and set the couter to 1. Thus, in order to keep the result in the calling function, you have to pass the matrix by reference.
If its true that you have to explicitly pass by reference, then how do all the OpenCV functions work? None of them pass values by reference, yet they somehow seem to write to the passed in Mat just fine. For example, here is the declaration for the Sobel function in imgproc.hpp:
//! applies generalized Sobel operator to the image
CV_EXPORTS_W void Sobel( InputArray src, OutputArray dst, int ddepth,
int dx, int dy, int ksize=3,
double scale=1, double delta=0,
int borderType=BORDER_DEFAULT );
as you can see, it passes in src and dst without a &. And yet I know that after I call the Sobel with an empty dst, it will end up filled. No '&' involved.