What is the difference between zero crossing and thresholding of an image apart from for zero crossing the threshold is zero? This is in the context of computer vision using a laplacian of Gaussian filter on an image.
The zero crossings form thin lines. A function will typically cross the zero (change signs) at a point.
Thresholding at zero would give you the areas where the function (image) is positive.
Zero crossings of the Laplacian indicate the locations of edges in the image. This is a noise-sensitive way of edge detection, useful more as a theoretical concept than a practical approach. Canny detects edges by looking for local maxima in the derivatives. That is a much more practical approach, though theoretically it's the same concept as finding zero crossings of the 2nd derivative (which the Laplacian is).