D = VL_LIOP(I) computes the LIOP descriptor of an image I, as described by [1]. I is a gray-scale square image with odd side length of class SINGLE. D is a column vector containing the LIOP descriptor of I. Note that LIOP is also integrated in the VL_COVDET() function for feature extraction.
VL_LIOP() accepts the following options:
Set the number of neighbours sampled to consruct the orer pattern of each image pixel.
Set the radius of the circular neighbourhood used to sampled the local order pattern of each pixel.
Set the number of spatial pooling regions. The LIOP descriptor has dimension factorial(NumNeighbours) * NumSpatialBins.
Set the intensity threshold used to weight oder patterns as they are pooled into a histogram. A negative value is interpreted as a fraction of the difference between the maximum and minimum intesity in each local patch.
If specified, be verbose
[1] Z. Wang, B. Fan, F. Wu. Local Intensity Order Pattern for feature description. In ICCV, 2011