Facial component extraction by cooperative active nets with global constraints

Parametric active membrane for segmentation of multiple objects in an image

“…There exists a set of surface evolution schemes where surface is designed as a net and most of them are targeted for specific applications [1,7,15,18]. In all these cases the surface is represented by a matrix of points fixed at regular spatial intervals.…”

“…In [7,18] the force is calculated from the probability distribution of the image pixels of the object to be segmented and therefore needs training samples. In [7] the external force is calculated from the hue and saturation distribution of the object color while in [18] external force is calculated from the color chromaticity diagram. However, [7,18] are not topology adaptive model.…”

“…In [7] the external force is calculated from the hue and saturation distribution of the object color while in [18] external force is calculated from the color chromaticity diagram. However, [7,18] are not topology adaptive model. In [15] multiple objects are detected using multiple initialization of the active net which increases the time complexity with the increase in the number of objects in the image.…”