Intel® Integrated Performance Primitives (Intel® IPP) Developer Guide and Reference
A newer version of this document is available. Customers should click here to go to the newest version.
Visible to Intel only — GUID: GUID-6B350139-F381-47C6-AEB4-7960963B7ABF
Visible to Intel only — GUID: GUID-6B350139-F381-47C6-AEB4-7960963B7ABF
Corner Detection
The Sobel and Scharr first derivative operators are to be used to take the x and y derivatives of an image. Then a small region of interest (ROI) is to be defined to detect corners in.
A 2x2 matrix of sums of the x and y derivatives is created as follows:

Solving

where λ is a column vector of the eigen values and I is the identity matrix, gives the eigen values. For the 2x2 matrix of the equation above, the solutions may be written in a closed form:

If λ1, λ2 > t, where t is some threshold, then a corner is considered to be found at that location. This can be very useful for object or shape recognition.