Most of today's image enhancement and gamut mapping algorithms are pixel based and can therefore suffer from severe clipping artifacts. We propose an algorithm that is based on super pixels, which is an (over-) segmentation of the input image. We tested various segmentation algorithms in combination with a number of gamut mapping algorithms and compared a selection of those in a perception experiment. The results show that for large enhancement the segmented approach is preferred, but for moderate enhancements this was not statistically significant. We also found that correcting for the Helmholtz-Kohlrausch effect did not improve the performance of the algorithm.