Democratic Tone Mapping Using Optimal K-means Clustering

The field of high dynamic range imaging addresses the problem of capturing and displaying the large range of luminance levels found in the world, using devices with limited dynamic range. In this paper we present a novel tone mapping algorithm that is based on \(K\)-means clustering. Using dynamic programming we are able to, not only solve the clustering problem efficiently, but also find the global optimum. Our algorithm runs in \(O(N^2K)\) for an image with \(N\) luminance levels and \(K\) output levels. We show that our algorithm gives comparable result to state-of-the-art tone mapping algorithms, but with the additional large benefit of a total lack of parameters. We test our algorithm on a number of standard high dynamic range images, and give qualitative comparisons to a number of state-of-the-art tone mapping algorithms.

[1]  Rae-Hong Park,et al.  Local tone mapping using K-means algorithm and automatic gamma setting , 2011, 2011 IEEE International Conference on Consumer Electronics (ICCE).

[2]  Paul Scheunders,et al.  A genetic c-Means clustering algorithm applied to color image quantization , 1997, Pattern Recognit..

[3]  Paul Scheunders,et al.  A comparison of clustering algorithms applied to color image quantization , 1997, Pattern Recognit. Lett..

[4]  Robert L. Stevenson,et al.  Dynamic range improvement through multiple exposures , 1999, Proceedings 1999 International Conference on Image Processing (Cat. 99CH36348).

[5]  Shree K. Nayar,et al.  High Dynamic Range from Multiple Images: Which Exposures to Combine?∗ , 2003 .

[6]  Rafal Mantiuk,et al.  Display adaptive tone mapping , 2008, SIGGRAPH 2008.

[7]  Werner Purgathofer,et al.  Tone Reproduction and Physically Based Spectral Rendering , 2002, Eurographics.

[8]  Mehmet Celenk,et al.  A color clustering technique for image segmentation , 1990, Comput. Vis. Graph. Image Process..

[9]  Sanjoy Dasgupta,et al.  Random projection trees for vector quantization , 2008, 2008 46th Annual Allerton Conference on Communication, Control, and Computing.

[10]  Richard Bellman,et al.  A note on cluster analysis and dynamic programming , 1973 .

[11]  Karol Myszkowski,et al.  Adaptive Logarithmic Mapping For Displaying High Contrast Scenes , 2003, Comput. Graph. Forum.

[12]  Éva Tardos,et al.  Algorithm design , 2005 .

[13]  Haizhou Wang,et al.  Ckmeans.1d.dp: Optimal k-means Clustering in One Dimension by Dynamic Programming , 2011, R J..

[14]  S. P. Lloyd,et al.  Least squares quantization in PCM , 1982, IEEE Trans. Inf. Theory.

[15]  Calle Lejdfors,et al.  Adaptive enhancement and noise reduction in very low light-level video , 2007, 2007 IEEE 11th International Conference on Computer Vision.

[16]  Alexei A. Efros,et al.  Fast bilateral filtering for the display of high-dynamic-range images , 2002 .

[17]  Holly E. Rushmeier,et al.  Tone reproduction for realistic images , 1993, IEEE Computer Graphics and Applications.

[18]  Hans-Peter Seidel,et al.  A perceptual framework for contrast processing of high dynamic range images , 2006, TAP.

[19]  Greg Ward,et al.  A Contrast-Based Scalefactor for Luminance Display , 1994, Graphics Gems.

[20]  Christine D. Piatko,et al.  A visibility matching tone reproduction operator for high dynamic range scenes , 1997 .

[21]  Jitendra Malik,et al.  Recovering high dynamic range radiance maps from photographs , 1997, SIGGRAPH.

[22]  E. Reinhard.  Photographic Tone Reproduction for Digital Images , 2002 .

[23]  Meena Mahajan,et al.  The Planar k-means Problem is NP-hard I , 2009 .

[24]  Douglas Steinley,et al.  K-means clustering: a half-century synthesis. , 2006, The British journal of mathematical and statistical psychology.

[25]  Magnus Oskarsson,et al.  The Remarkable Visual Abilities of Nocturnal Insects: Neural Principles and Bioinspired Night-Vision Algorithms , 2014, Proceedings of the IEEE.

[26]  Pierre Hansen,et al.  NP-hardness of Euclidean sum-of-squares clustering , 2008, Machine Learning.