As written in title, i want to solve Ax=b.
When creating A with cv::Mat and my needed dimensions of (10 million x 5 million entries) the allocation of this storage will (obviously) fail :-D.
Instead I want to use cv::SparseMat cause every row has only two entries (1 or -1) this should be the right way to go.
But how is it possible to solve Ax=b using OpenCV and SparseMat?
With cv::Mat this is done using cv::solve(A,b,x, Solver) but it wont work (its not a correct input argument of type cv::InputArray) with cv::SparseMat any ideas how I can solve x using OpenCV (3.1.0)?
Related
I'm using the OpenCV library in a CUDA C++ environment to resize an image obtained by GPU processing.
A crucial step of the processing involves resampling the image and INVERTING the aspect ratio.
Example problem:
Resize a 2000 x 500 and transform it into a 500 x 2000 image using CUDA
This is attempted by the following OpenCV command:
cv::cuda::resize(d_src,d_dst,cv::Size(500,2000),cv::INTER_CUBIC);
Where d_src and d_dst are the proper GpuMats with 2000 x 500 and 500 x 2000 size.
The maximum permitted resize is a square of either 2000x2000 or 500x500. The function behaves as expected as long as the aspect ratio is not inverted. I have also attempted making the interpolation in two steps, either by expansion and reduction:
Going from 2000x500 to 2000x2000 to 500x2000.
cv::cuda::resize(d_src,d_buffer,cv::Size(2000,2000),cv::INTER_CUBIC);
cv::cuda::resize(d_buffer,d_dst,cv::Size(500,2000),cv::INTER_CUBIC);
Going from 2000x500 to 500x500 to 500x2000.
cv::cuda::resize(d_src,d_buffer,cv::Size(500,500),cv::INTER_CUBIC);
cv::cuda::resize(d_buffer,d_dst,cv::Size(500,2000),cv::INTER_CUBIC);
Both these approaches fail and are not preferred since they consume a considerable amount of extra GPU memory.
Has anyone experienced a similar problem with this function? Could somebody help me out?
Thank you in advance
Nevermind. This problem can be solved by setting the size parameters to auto:
cv::cuda::resize(d_src,d_dst,d_dst.size(),0,0,cv::INTER_CUBIC);
I've been trying to refine my camera parameters with CvLevMarq but after reading about it, it seems to be causing mixed results - which is exactly what I am experiencing. I read about the alternatives and came upon EIGEN - and also found this library that utilizes it.
However, the library above seems to use a stitching class that doesn't support OpenCV and will probably require me to port it to OpenCV.
Before going ahead and doing so, which will probably not be an easy task, I figured I'd ask around first and see if anyone else had the same problem?
I'm currently using:
1. Calculating features with FASTFeatureDetector
Ptr<FeatureDetector> detector = new FastFeatureDetector(5,true);
detector->detect(firstGreyImage, features_global[firstImageIndex].keypoints); // Previous picture
detector->detect(secondGreyImage, features_global[secondImageIndex].keypoints); // New picture
2. Extracting features with SIFTDescriptorExtractor
Ptr<SiftDescriptorExtractor> extractor = new SiftDescriptorExtractor();
extractor->compute(firstGreyImage, features_global[firstImageIndex].keypoints, features_global[firstImageIndex].descriptors); // Previous Picture
extractor->compute(secondGreyImage, features_global[secondImageIndex].keypoints, features_global[secondImageIndex].descriptors); // New Picture
3. Matching features with BestOf2NearestMatcher
vector<MatchesInfo> pairwise_matches;
BestOf2NearestMatcher matcher(try_use_gpu, 0.50f);
matcher(features_global, pairwise_matches);
matcher.collectGarbage();
4. CameraParams.R quaternion passed from a device (slightly inaccurate which causes the issue)
5. CameraParams.Focal == 389.0f -- Played around with this value, 389.0f is the only value that matches the images horizontally but not vertically.
6. Bundle Adjustment (cvLevMarq, calcError & calcJacobian)
Ptr<BPRefiner> adjuster = new BPRefiner();
adjuster->setConfThresh(0.80f);
adjuster->setMaxIterations(5);
(*adjuster)(features,pairwise_matches,cameras);
7. ExposureCompensator (GAIN)
8. OpenCV MultiBand Blender
What works so far:
SeamFinder - works to some extent but it depends on the result of the cvLevMarq algoritm. I.e. if the algoritm is off, seamFinder is going to be off too.
HomographyBasedEstimator works beautifully. However, since it "relies" on the features, it's unfortunately not the method that I'm looking for.
I wouldn't want to rely on the features since I already have the matrix, if there's a way to "refine" the current matrix instead - then that would be the targeted result.
Results so far:
cvLevMarq "Russian roulette" 6/10:
This is what I'm trying to achieve 10/10 times. But 4/10 times, it looks like the picture below this one.
By simply just re-running the algorithm, the results change. 4/10 times it looks like this (or worse):
cvLevMarq "Russian roulette" 4/10:
Desired Result:
I'd like to "refine" my camera parameters with the features that I've matched - in hope that the images would align perfectly. Instead of hoping that cvLevMarq will do the job for me (which it won't 4/10 times), is there another way to ensure that the images will be aligned?
Update:
I've tried these versions:
OpenCV 3.1: Using CVLevMarq with 3.1 is like playing Russian roulette. Some times it can align them perfectly, and other times it estimates focal as NAN which causes segfault in the MultiBand Blender (ROI = 0,0,1,1 because of NAN)
OpenCV 2.4.9/2.4.13: Using CvLevMarq with 2.4.9 or 2.4.13 is unfortunately the same thing minus the NAN issue. 6/10 times it can align the images perfectly, but the other 4 times it's completely off.
My Speculations / Thoughts:
Template Matching using OpenCV. Maybe if I template match the ends of the images (i.e. x = 0, y = 0,height = image.height, width = 50). Any thoughts about this?
I found this interesting paper about Levenberg Marquardt applied in Homography. That looks like something that could solve my problem since the paper uses corner detection and whatnot to detect the features in the images. Any thoughts about this?
Maybe the problem isn't in CvLevMarq but instead in BestOf2NearestMatcher? However, I've searched for days and I couldn't find another method that returns the pairwise matches to pass to BPRefiner.
Hough Line Transform Detecting the lines in the first/second image and use that to align the images. Any thoughts on this? -- One thing might be, what if the images doesn't have any lines? I.e. empty wall?
Maybe I'm overkilling something so simple.. Or maybe I'm not? Basically, I'm trying to align a set of images so I can warp them without overlapping each-other. Drop a comment if it doesn't make sense :)
Update Aug 12:
After trying all kinds of combinations, the absolute best so far is CvLevMarq. The only problem with it is the mixed results shown in the images above. If anyone has any input, I'd be forever grateful.
It seems your parameter initialization is the problem. I would use a linear estimator first, i.e. ignore your noisy sensor, and then use this as the initial values for the non-linear optimizer.
A quick method is to use getaffinetransform, as you have mostly rotation.
Maybe you want to take a look at this library: https://github.com/ethz-asl/kalibr.
Cheers
If you want to stitch the images, you should see stitching_detailed.cpp. It will probably solve your problem.
In addition, I have used Graph Cut Seam Finding method with Canny Edge Detection for better stitching results in this code. If you want to optimize this code, see here.
Also, if you are going to use it for personal use, SIFT is good. You should know, SIFT is patented and will cost you if you use it for commercial purposes. Use ORB instead.
Hope it helps!
I am working on an algorithm with many computations done on a GPU. I'm working mainly with oclMat structures and am trying to avoid copying from CPU to GPU and vice versa, yet I cannot find an easy way to:
compare all elements in an ocl matrix to a specific single value (be it float or double, for instance) and create a logical matrix in accordance
create an oclMat matrix with a given size and type initialized with all elements to a specific value (for example all elements are float and equal to 1.234567)
For example:
cv::ocl::oclMat M1 =...
// DO STUFF WITH M1
cv::ocl::oclMat logicalM1 = M1>1.55; // compare directly to a single value
cv::ocl::oclMat logicalM2 = ... ; // i.e. I want a 100x100 CV_32FC1 matrix with all elements set to be equal to 1.234567
By reading the documentation, it seems using cv::ocl::compare only works with both matrices the same dimensions and type, so maybe my first request isn't feasible. On the other hand, I don't know how to initialize a specific matrix directly in ocl (with cv::Mat I know how it's done).
I assume an easy workaround exists, but haven't found one yet... Thanks!
You are right. Looks like cv::ocl::compare supports only two cv::oclMat on input.
But you can create oclMat filled with specific value as follows:
cv::ocl::oclMat logicalM2(M1.size(), M1.type);
logicalM2.setTo(cv::Scalar(1.234567));
cv::ocl::oclMat logicalM1;
cv::ocl::compare(M1, logicalM2, logicalM1, cv::CMP_GT);
P.S. Also I suggest you trying new OpenCV 3.0 with Transparent-API which makes processing on GPU using OpenCL much easier.
I have mat with only one column and 1600 rows. I want to filter it using a Gaussian.
I tried the following:
Mat AFilt=Mat(palm_contour.size(),1,CV_32F);
GaussianBlur(A,AFilt,cv::Size(20,1),3);
But I get the exact same values in AFilt (the filtered mat) and A. It looks like GaussianBlur has done nothing.
What's the problem here? How can I smooth a single-column mat with a Gaussian kernel?
I read about BaseColumnFilt, but haven't seen any usage examples so I'm not sure how to use them.
Any help given will be greatly appreciated as I don't have a clue.
I'm working with OpenCV 2.4.5 on windows 8 using Visual Studio 2012.
Thanks
Gil.
You have a single column but you are specifying the width of the gaussian to be big instead of specifying the height! OpenCV use row,col or x,y notation depending on the context. A general rule is whenever you use Point or Size, they behave like x,y and whenever the parameters are separate values they behave like row,col.
The kernel size should also be odd. If you specify the kernel size you can set sigma to zero to let OpenCV compute a suitable sigma value.
To conclude, this should work better:
GaussianBlur(A,AFilt,cv::Size(1,21),0);
The documentation og GaussianBlur says the kernel size must be odd, I would try using an odd size kernel and see if that makes any difference
I am trying to do a 2D Real To Complex FFT using CUFFT.
I realize that I will do this and get W/2+1 complex values back (W being the "width" of my H*W matrix).
The question is - what if I want to build out a full H*W version of this matrix after the transform - how do I go about copying some values from the H*(w/2+1) result matrix back to a full size matrix to get both parts and the DC value in the right place
Thanks
I'm not familiar with CUDA, so take that into consideration when reading my response. I am familiar with FFTs and signal processing in general, though.
It sounds like you start out with an H (rows) x W (cols) matrix, and that you are doing a 2D FFT that essentially does an FFT on each row, and you end up with an H x W/2+1 matrix. A W-wide FFT returns W values, but the CUDA function only returns W/2+1 because real data is even in the frequency domain, so the negative frequency data is redundant.
So, if you want to reproduce the missing W/2-1 points, simply mirror the positive frequency. For instance, if one of the rows is as follows:
Index Data
0 12 + i
1 5 + 2i
2 6
3 2 - 3i
...
The 0 index is your DC power, the 1 index is the lowest positive frequency bin, and so forth. You would thus make your closest-to-DC negative frequency bin 5+2i, the next closest 6, and so on. Where you put those values in the array is up to you. I would do it the way Matlab does it, with the negative frequency data after the positive frequency data.
I hope that makes sense.
There are two ways this can be acheived. You will have to write your own kernel to acheive either of this.
1) You will need to perform conjugate on the (half) data you get to find the other half.
2) Since you want full results anyway, it would be best if you convert the input data from real to complex (by padding with 0 imaginary) and performing the complex to complex transform.
From practice I have noticed that there is not much of a difference in speed either way.
I actually searched the nVidia forums and found a kernel that someone had written that did just what I was asking. That is what I used. if you search the cuda forum for "redundant results fft" or similar you will find it.