Loss of Class fisheye correcting wide-angle camera in OpenCV 3.0

asked 2016-09-20 07:48:54 -0600

Hello everyone, I'm trying to find a method to stitching two images which are from two fish-eye camera. But I have no idea how to correct the original images. I have tried to do this with class fisheye in opencv3.0 but it has a big loss of view so that I have no enough information to stitch in next step. So is there any idea to solve this problem? Or should I adapt other models and algorithms? I‘m fortunate if you could show me a specific strategy. Thank you very much!

edit retag flag offensive close merge delete


Have you calibrated your camera? What is the input and output image that you are getting now? What are the parameters you used for the output camera matrix?

Balaji R gravatar imageBalaji R ( 2016-09-21 03:57:56 -0600 )edit

Thanks for your reply! I have got 43 images to calibrate my camera. Here are the parameters I got Camera Matrix: [223.6587415330009, 0, 290.8699580366777; 0, 259.9323970375434, 309.1011658440989; 0, 0, 1]

Dist Coefficients: [-0.0157997, -0.00707966, 0.00217689, -0.000716948]

The input and output images are shown as follow: link text

You can see there is a big loss of view. It can't provide enough information for stitching. Do you have any idea to solve this problem? Thank your very much!

damingxing gravatar imagedamingxing ( 2016-09-21 04:27:01 -0600 )edit