Finding tires on a car.

I have been working on extracting information from a picture of a car. I am having trouble finding the tires on this car. I have them down to distinct circles in a BW binary image. I have tried several codes and methods online using hough transforms but with no success. When I try Tao's code: http://www.mathworks.com/matlabcentral/fileexchange/9168 I get an error. My grayscale image is not 2-D. Any suggestions on how to convert it? Thanks for your help in advance.

Answer 1

[Walter Roberson]

Does ndims(YourImage) show 3 ? If so and it is a grayscale image, YourImage(:,:,1) would be appropriate for processing as a 2D grayscale image.

Answer 2

[David Young]

Also see rgb2gray from the Image Processing Toolbox.

Note that you should not binarise your image before using Tao's code - it expects the original image, if necessary converted to grayscale.

[Walter Roberson]

Though if the image is an RGB image that appears as grayscale, then just selecting the single plane will have the same effect as rgb2gray()

[David Young]

Walter: yes, indeed.

Answer 3

[Zach Chipman]

Ok I finally found a much more simple way to do this. I used regionprops.eccentricity and compared some other known information about the object to see just how circular it was. Thank you for all of your advice!