Geek asks and answers

Detect paper corners with iOS 5 AV Foundation and real-time core-image

I am currently creating a prototype camera application that should recognize sheets of paper lying on a table. The hint about this is that it should recognize in real time, so I grab the camera’s video stream, which in iOS 5 can easily be done using the AV framework. I looked here and here

They perform basic object recognition there.

I found that using the OpenCV library in this real-time environment does not work for real.

So I need an algorithm to determine the edges of an image without OpenCV.

Does anyone have any code snippets that lay out how to do this or direct me in the right direction.

Any help would be appreciated.

+3
source share
2 answers

You will not be able to do this with the current implementation of Core Image on iOS, because angle detection requires some operations that Core Image does not yet support. However, I am developing an open source framework called GPUImage , which has the necessary capabilities.

To find the corners of an object, you can use the accelerated GPU implementation of the Harris angle detection algorithm, which I just started working. You may need to adjust the thresholds, sensitivity and size of the input image to work in your specific application, but it can return the corners for pieces of paper that it finds in the scene:

Harris corners

, , , .

, , , GPUImage GitHub GPUImageHarrisCornerDetectionFilter GPUImageVideoCamera. , .

iPhone 4 ~ 15-20 FPS 640x480, , , ~ 10 FPS. , . IPhone 4S 20-25 FPS, . , .

+10

Brad GPUImage , , .

- :

  • .

@Mirco - ?

@Brad. , , 5 ? ...

+2

More articles:


All Articles