Finds edges in an image using the Canny algorithm

edges = cv.Canny(image, thresh)
edges = cv.Canny(image, thresh, 'OptionName', optionValue, ...)

Input

• image 8-bit input image (grayscale or color image).
• thresh Threshold for the hysteresis procedure. Scalar or 2-element vector [low_thresh,high_thresh].

Output

• edges Output edge map; single channels 8-bit image, which has the same size as image.

Options

• ApertureSize Aperture size for the Sobel operator. Default 3
• L2Gradient Flag indicating whether a more accurate L2 norm sqrt((dI/dx)^2 + (dI/dy)^2) should be used to compute the image gradient magnitude (L2gradient=true), or whether the default L1 norm abs(dI/dx) + abs(dI/dy) is enough (L2gradient=false). Default false

The function finds edges in the input image image and marks them in the output map edges using the Canny algorithm. When thresh is 2-element vector, the smallest value between them is used for edge linking. The largest value is used to find initial segments of strong edges. When thresh is a scalar, it is treated as a higher threshold value and 0.4*thresh is used for the lower threshold. See http://en.wikipedia.org/wiki/Canny_edge_detector

References

[Canny86]:

John Canny. A computational approach to edge detection. Pattern Analysis and Machine Intelligence, IEEE Transactions on, (6):679-698, 1986.