<div dir="ltr">Hi Igor,<div><br></div><div>This is something that connected component labelling can do fairly well. The procedure is as follows (roughly):</div><div><br></div><div>1) Find the brightness of the voxels within the screw. Find the brightness of the voxels surrounding the screw. Choose a threshold that separates them. Apply that threshold to the image to make a binary image.</div><div><br></div><div>2) Apply a 3D connected component labelling filter to label the objects in the binary image.</div><div><br></div><div>3) Identify the object that is most likely to be the screw, based on size, shape, or other characteristics.</div><div><br></div><div>See the following website for a quick description of connected component labelling (the site describes it in 2D, but the methods works just as well in 3D):</div><div><a href="http://homepages.inf.ed.ac.uk/rbf/HIPR2/label.htm">http://homepages.inf.ed.ac.uk/rbf/HIPR2/label.htm</a><br></div><div><br></div><div>ITK has a ConnectedComponentImageFilter. For VTK, I have a vtkImageConnectivityFilter in one of my own repositories:</div><div><a href="https://github.com/dgobbi/AIRS/blob/master/ImageSegmentation/vtkImageConnectivityFilter.h">https://github.com/dgobbi/AIRS/blob/master/ImageSegmentation/vtkImageConnectivityFilter.h</a></div><div><br></div><div> - David</div><div><br><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Nov 11, 2014 at 4:52 AM, igorznt <span dir="ltr"><<a href="mailto:igorznt@gmail.com" target="_blank">igorznt@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">I have to recognize a screw in a CT an locate the point on its bases and<br>
head. Does anyone knows how or could indicate a good material about?<br></blockquote></div></div></div></div>