Menu
I am trying to find a good algorihtm that would detect corners in a image in a mobile phone. There are multiple algorithms to do that I am not sure which one will perform better in a memory and processor limited environment.
Specifically I am trying to find a sudoku grid in a picture taken using the camera of the phone. I am using C# and could not find any libraries that has basic image processing features. I implemented a Sobel filter to do edge detection and that is where I stand.
To make it clear the question is does anybody have any suggestions to use a specific algorithm or a library?
528
The Harris corner detector works by taking horizontal and vertical derivatives of the image and looking for areas where both are high, this is quantified by the Harris corner descriptor which is defined in our case as the matrix �and the descriptor is .
The Harris corner detector is a corner detection operator that is commonly used in computer vision algorithms to extract corners and infer features of an image.
Shi-Tomasi is a corner detector that is used to extract strongest corners from images and other corners below level are rejected. It uses minimum euclidean distance between corners detected. Then sorts the remaining corners based on quality in descending order.
Features from accelerated segment test (FAST) is a corner detection method, which could be used to extract feature points and later used to track and map objects in many computer vision tasks. The FAST corner detector was originally developed by Edward Rosten and Tom Drummond, and was published in 2006.
I wouldn't say "corner detection" by itself is a very good way to do this. Take a step back and think about a photo of a sodoku grid, there are probably lots of assumptions you can make to simplify things.
For example, a sodoku grid always looks exactly the same:
treating the image in the HSV colour space will allow you to look for high lightness areas (white-ish colours), RGB is a bit pants for most image-processing techniques.
thresholding the image should then reduce noise
Adjusting the image histogram first may also give you better results as it will probably whiten the grid (depends on the image though).
Then all you have to do is find a square. Because you know the grid is regular within it, you can divide the pixels up accordingly and OCR the squares with a number in.
:D
172
Since you are looking for a regular 9x9 grid consider the Hough transform. One way is to run an edge detector first, find all straight lines using the original Hough transform, and then try to figure out which of them form a grid. Or maybe you can come up with a clever way to parametrize the whole grid.
35
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With
