I wanted to capture a pair of stereo images with my iPhone camera. I wanted to know how could I do it. What ai have tried so far is, clicked one photo moved my phone a little to the right (almost 1.5 inches), but I did not get the desired results.
I want to rectify those images before I could apply opencv's StereoBM_create function to them. But I can't capture proper stereo images.
Related
I am trying to organize my trading card collection digitally and am working on building a scanner using ocr to detect the names of my collection.
I need to use a webcam to snap a single image of each card in question. Snapping the image doesn't seem to be to difficult, but I need help determining how to get OpenCV to capture only a specific part of that image for OCR to work with. I'm trying to capture just the text portion of the image so that the artwork on the cards doesn't interfere with the OCR.
If my card will be placed in the same physical location each time, is there a way to get OpenCV to take an image and focus on just the area of the image that I'm interested in.
Thank You
Sour Jack
I am not sure I understand the problem. Do you want to use your OCR algorithm always on the same portion of the snapshot? If so, you can try something like:
roi = img[y:y+height, x:x+width]
There is more information here: http://answers.opencv.org/question/29260/how-to-save-a-rectangular-roi/
Using Raspberry Pi NoIR camera, I am trying to do occupancy detection using OpenCV for shared office space.
In order to identify vacant spots, I am using image of empty office space as a background model. Once we have the background model, I subtract the live video frame from the background image model(using cvAbsDiff) and identify the changes.
I am getting the images using a webcam mounted on the roof. The problem is to identify a person from this angle. When I work with live video, there are subtle variations (changes in chair position, physical objects or variations of camera) which act as noise. I tried using BackgroundSubtractorMOG, MOG2 but MOG and MOG2 can only do differences from a video stream but not using an image reference model.
Any inputs on possible directions will be greatly appreciated. I will be happy to share the technique with you guys that works once I have a working solution. Thanks!
Some background:
Hi all! I have a project which involves cloud imaging. I take pictures of the sky using a camera mounted on a rotating platform. I then need to compute the amount of cloud present based on some color threshold. I am able to this individually for each picture. To completely achieve my goal, I need to do the computation on the whole image of the sky. So my problem lies with stitching several images (about 44-56 images). I've tried using the stitch function on all and some subsets of image set but it returns an incomplete image (some images were not stitched). This could be because of a lack of overlap of something, I dunno. Also the output image has been distorted weirdly (I am actually expecting the output to be something similar to a picture taken by a fish-eye lense).
The actual problem:
So now I'm trying to figure out the opencv stitching pipeline. Here is a link:
http://docs.opencv.org/modules/stitching/doc/introduction.html
Based on what I have researched I think this is what I want to do. I want to map all the images to a circular shape, mainly because of the way how my camera rotates, or something else that has uses a fairly simple coordinate transformation. So I think I need get some sort of fixed coordinate transform thing for the images. Is this what they call the homography? If so, does anyone have any idea how I can go about my problem? After this, I believe I need to get a mask for blending the images. Will I need to get a fixed mask like the one I want for my homography?
Am I going through a possible path? I have some background in programming but almost none in image processing. I'm basically lost. T.T
"So I think I need get some sort of fixed coordinate transform thing for the images. Is this what they call the homography?"
Yes, the homography matrix is the transformation matrix between an original image and the ideal result. It warps an image in perspective so it can fit in stitching to the other image.
"If so, does anyone have any idea how I can go about my problem?"
Not with the limited information you provided. It would ease the problem a lot if you know the order of pictures (which borders which.. row, column position)
If you have no experience in image processing, I would recommend you use a tutorial covering stitching using more basic functions in detail. There is some important work behind the scenes, and it's not THAT harder to actually do it yourself.
Start with this example. It stitches two pictures.
http://ramsrigoutham.com/2012/11/22/panorama-image-stitching-in-opencv/
I am working on a project with gesture recognition. Now I want to prepare a presentation in which I can only show images. I have a series of images defining a gesture, and I want to show them in a single image just like motion history images are shown in literature.
My question is simple, which functions in opencv can I use to make a motion history image using lets say 10 or more images defining the motion of hand.
As an example I have the following image, and I want to show hand's location (opacity directly dependent on time reference).
I tried using GIMP to merge layers with different opacity to do the same thing, however the output is not good.
You could use cv::updateMotionHistory
Actually OpenCV also demonstrates the usage in samples/c/motempl.c
I'm building a web cam application as my C++ project in my college. I am integrating QT (for GUI) and OpenCV (for image processing). My application will be a simple web cam app that will access the web cam, show/record videos, capture images and other stuffs.
Well, I also want to put in a feature to add cliparts to captured images, or the streaming video. While on my research, I found out that there is no way we can overlay two images using OpenCV. The best alternative I was able to find was to reconfigure the whole image to add the clipart into the original image making it a single image. You see, that's not going to work for me as I have to be able to move the clipart and resize or rotate the clipart in my canvas.
So, I was wondering if anybody could tell me how to achieve the effect I want most efficiently.
I would really appreciate your help. The deadline for the project submission is closing in and its a huge bump on the road to completion. PLEEEASE... RELP!!
If you just want to stick a logo onto the openCV image then you simply define a region of interest (roi) on the destination video image and copy the source image to this (the details vary with each version of opencv)
If you want the logo to be semi transparent - like a TV channel ID - then you can copy the image but loop over the pixels writing a destination that is source_pixel/2 + dest_pixel/2;