xpete's profile - activity

I have an image that is 90% equal but 10% or more can change like you said. Image you have an image with 4 quadrants and 2 are switched from the original but it still recognizes if it was the original image.

thanks for the replies. pixel by pixel comparison is too slow.

Good morning. We’re working in a computer vision Project, running on mobile devices with android operating systems; the goal of the application is to detect (through the device’s camera) if a specific image is exactly the same as the reference image. We are using the OpenCV library and following the tips on this book. we’ve already made several experiences using “STAR”, “FREAK” and “BRUTE FORCE HAMMING” and we are still stumbling on a huge obstacle: the application recognizes/ accepts the correct image, but also a similar image (with little variations). We need that the application recognizes ONLY the exact image. Do you have any suggestions?

I'am trying to use OpenCV with libgdx to process the image from an Android phone camera and use it on a game.

So far I've a CvCameraViewListener2 implementation which is called from the GameScreen but the screen is always black. I'am not getting any imagem from the camera or any error.

I already did this: https://github.com/libgdx/libgdx/wiki/Integrating-libgdx-and-the-device-camera

I also wrote the OpenCV processing on a diferent App and it's working... now I want to merge that with LibGDX but it's not working.

Already tried on Nexus S, Galaxy S and Huawei u8815. All with stock software.

If I change mNumCameras to 1 sometimes it works... but not always.

Code is available here: https://github.com/xpete/libgdxopencvtest

Any help?

Thanks