IMAGE CONTENT RETARGETING by Alessandro Artusi Tunç Ozan Aydın Olga Sorkine-Hornung Francesco Banterle & Daniele Panozzo

Author:Alessandro Artusi, Tunç Ozan Aydın, Olga Sorkine-Hornung, Francesco Banterle & Daniele Panozzo
Language: eng
Format: epub
Publisher: CRC Press

Figure 4.17. An example of HaCohen et al.’s method [106]: (a) A source image. (b) A target image. (c) Found dense matches between (a) and (b). (d) The color transform; a spline. (e) The image in (a) after applying the color transform (d); note that colors are matched with the ones in (b).

Histogram-based approaches. Histograms can also be exploited in large collections; these are computationally cheap to compute and manipulate. Moreover, histograms of such photographs, i.e., the same scene, can exhibit similar color distributions.

A straightforward method is to align histograms of photographs for each color channel [245]. The alignment of histograms requires extraction of features for each histogram; which are typically local maxima computed in scale space [160] to reduce outliers and noise influence. Given two gray-scale images, I1 and I2, their respective histograms are computed, h1 and h2. Then, feature vectors are extracted, and , and matched using the minimum Euclidean distance. After this step, the color mapping, f, is a polynomial whose coefficients are found by minimizing an energy term, E, defined as

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.