A method of
gamut mapping maps an input image composed of pixels and having an input
gamut (IG) defined by input RGB primaries (Ri, Gi, Bi) to a
reproduction gamut (RG) defined by
reproduction RGB primaries (Ro, Go, Bo). The
reproduction gamut (RG) is narrower than the input gamut (IG). An input
signal (RGBin) defined with respect to the input RGB primaries (Ri, Gi, Bi) is color transformed (1) into a transformed
signal (RGBt) defined with respect to the reproduction RGB primaries (Ro, Go, Bo), whereby color information of the pixels (Pl, P2, P3) within the reproduction gamut (RG) is preserved. Scaling factors (SFi) indicating a distance between on the one hand pixels (Pl, P2, P3) of the transformed
signal (RGBt) which are outside the reproduction gamut (RG), and on the other hand an edge of the reproduction gamut (RG) are determined (2). The scaling factors (SFi) of pixels (Pl, P2, P3) of the transformed signal (RGBt) are spatial low-pass filtered (3) to obtain filtered factors (FFi) for these pixels (Pl, P2, P3). Component signal values of the transformed signal (RGBt) of the pixels (Pl, P2, P3) are corrected (4) in response to the filtered factors (FFi) to obtain reproduction signal values (RGBo) which are input values (RGBi) moved towards inside the reproduction gamut (RG).