Exploiting Local and Global Patch Rarities for Saliency Detection

Exploiting Local and Global Patch Rarities for Saliency Detection. Borji, A. & Itti, L. In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, Rhode Island, pages 1-8, Jun, 2012.
abstract bibtex

We introduce a saliency model based on two key ideas. The first one is considering local and global image patch rarities as two complementary processes. The second one is based on our observation that for different images, one of the RGB and Lab color spaces outperforms the other in saliency detection. We propose a framework that measures patch rarities in each color space and combines them in a final map. For each color channel, first, the input image is partitioned into non-overlapping patches and then each patch is represented by a vector of coefficients that linearly reconstruct it from a learned dictionary of patches from natural scenes. Next, two measures of saliency (Local and Global) are calculated and fused to indicate saliency of each patch. Local saliency is distinctiveness of a patch from its surrounding patches. Global saliency is the inverse of a patch’s probability of happening over the entire image. The final saliency map is built by normalizing and fusing local and global saliency maps of all channels from both color systems. Extensive evaluation over four benchmark eye-tracking datasets shows the significant advantage of our approach over 10 state-of-the-art saliency models.

@inproceedings{ Borji_Itti12cvpr,
  author = {A. Borji and L. Itti},
  title = {Exploiting Local and Global Patch Rarities for Saliency Detection},
  booktitle = {Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, Rhode Island},
  abstract = {We introduce a saliency model based on two key ideas.  The first one is considering local and global image
                  patch rarities as two complementary processes. The second one is based on our observation that for
                  different images, one of the RGB and Lab color spaces outperforms the other in saliency detection. We
                  propose a framework that measures patch rarities in each color space and combines them in a final
                  map. For each color channel, first, the input image is partitioned into non-overlapping patches and
                  then each patch is represented by a vector of coefficients that linearly reconstruct it from a learned
                  dictionary of patches from natural scenes. Next, two measures of saliency (Local and Global) are
                  calculated and fused to indicate saliency of each patch. Local saliency is distinctiveness of a patch
                  from its surrounding patches. Global saliency is the inverse of a patch’s probability of happening
                  over the entire image.  The final saliency map is built by normalizing and fusing local and global
                  saliency maps of all channels from both color systems. Extensive evaluation over four benchmark
                  eye-tracking datasets shows the significant advantage of our approach over 10 state-of-the-art
                  saliency models.},
  pages = {1-8},
  month = {Jun},
  year = {2012},
  review = {full/conf},
  type = {bu;mod},
  if = {2012 acceptance rate: 26.2%},
  file = {http://ilab.usc.edu/publications/doc/Borji_Itti12cvpr.pdf}
}

Downloads: 0

{"_id":{"_str":"5298a1a19eb585cc2600091f"},"__v":0,"authorIDs":[],"author_short":["Borji, A.","Itti, L."],"bibbaseid":"borji-itti-exploitinglocalandglobalpatchraritiesforsaliencydetection-2012","bibdata":{"html":"<div class=\"bibbase_paper\"> \n\n\n<span class=\"bibbase_paper_titleauthoryear\">\n\t<span class=\"bibbase_paper_title\"><a name=\"Borji_Itti12cvpr\"> </a>Exploiting Local and Global Patch Rarities for Saliency Detection.</span>\n\t<span class=\"bibbase_paper_author\">\nBorji, A.; and Itti, L.</span>\n\t\n</span>\n\n\n\nIn\n<i>Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, Rhode Island</i>, page 1-8, Jun 2012.\n\n\n\n\n\n<br class=\"bibbase_paper_content\"/>\n\n<span class=\"bibbase_paper_content\">\n \n \n \n <a href=\"javascript:showBib('Borji_Itti12cvpr')\"\n class=\"bibbase link\">\n \n\t\n\t\n\t\n BibTeX\n <i class=\"fa fa-caret-down\"></i></a>\n \n \n  \n <a class=\"bibbase_abstract_link bibbase link\"\n href=\"javascript:showAbstract('Borji_Itti12cvpr')\">\n Abstract\n <i class=\"fa fa-caret-down\"></i></a>\n \n \n \n\n \n \n \n</span>\n\n<div class=\"well well-small bibbase\" id=\"bib_Borji_Itti12cvpr\"\n style=\"display:none\">\n <pre>@inproceedings{ Borji_Itti12cvpr,\n author = {A. Borji and L. Itti},\n title = {Exploiting Local and Global Patch Rarities for Saliency Detection},\n booktitle = {Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, Rhode Island},\n abstract = {We introduce a saliency model based on two key ideas. The first one is considering local and global image\n patch rarities as two complementary processes. The second one is based on our observation that for\n different images, one of the RGB and Lab color spaces outperforms the other in saliency detection. We\n propose a framework that measures patch rarities in each color space and combines them in a final\n map. For each color channel, first, the input image is partitioned into non-overlapping patches and\n then each patch is represented by a vector of coefficients that linearly reconstruct it from a learned\n dictionary of patches from natural scenes. Next, two measures of saliency (Local and Global) are\n calculated and fused to indicate saliency of each patch. Local saliency is distinctiveness of a patch\n from its surrounding patches. Global saliency is the inverse of a patch’s probability of happening\n over the entire image. The final saliency map is built by normalizing and fusing local and global\n saliency maps of all channels from both color systems. Extensive evaluation over four benchmark\n eye-tracking datasets shows the significant advantage of our approach over 10 state-of-the-art\n saliency models.},\n pages = {1-8},\n month = {Jun},\n year = {2012},\n review = {full/conf},\n type = {bu;mod},\n if = {2012 acceptance rate: 26.2%},\n file = {http://ilab.usc.edu/publications/doc/Borji_Itti12cvpr.pdf}\n}</pre>\n</div>\n\n\n<div class=\"well well-small bibbase\" id=\"abstract_Borji_Itti12cvpr\"\n style=\"display:none\">\n We introduce a saliency model based on two key ideas. The first one is considering local and global image patch rarities as two complementary processes. The second one is based on our observation that for different images, one of the RGB and Lab color spaces outperforms the other in saliency detection. We propose a framework that measures patch rarities in each color space and combines them in a final map. For each color channel, first, the input image is partitioned into non-overlapping patches and then each patch is represented by a vector of coefficients that linearly reconstruct it from a learned dictionary of patches from natural scenes. Next, two measures of saliency (Local and Global) are calculated and fused to indicate saliency of each patch. Local saliency is distinctiveness of a patch from its surrounding patches. Global saliency is the inverse of a patch’s probability of happening over the entire image. The final saliency map is built by normalizing and fusing local and global saliency maps of all channels from both color systems. Extensive evaluation over four benchmark eye-tracking datasets shows the significant advantage of our approach over 10 state-of-the-art saliency models.\n</div>\n\n\n</div>\n","downloads":0,"bibbaseid":"borji-itti-exploitinglocalandglobalpatchraritiesforsaliencydetection-2012","role":"author","year":"2012","type":"bu;mod","title":"Exploiting Local and Global Patch Rarities for Saliency Detection","review":"full/conf","pages":"1-8","month":"Jun","key":"Borji_Itti12cvpr","if":"2012 acceptance rate: 26.2%","id":"Borji_Itti12cvpr","file":"http://ilab.usc.edu/publications/doc/Borji_Itti12cvpr.pdf","booktitle":"Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, Rhode Island","bibtype":"inproceedings","bibtex":"@inproceedings{ Borji_Itti12cvpr,\n author = {A. Borji and L. Itti},\n title = {Exploiting Local and Global Patch Rarities for Saliency Detection},\n booktitle = {Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, Rhode Island},\n abstract = {We introduce a saliency model based on two key ideas. The first one is considering local and global image\n patch rarities as two complementary processes. The second one is based on our observation that for\n different images, one of the RGB and Lab color spaces outperforms the other in saliency detection. We\n propose a framework that measures patch rarities in each color space and combines them in a final\n map. For each color channel, first, the input image is partitioned into non-overlapping patches and\n then each patch is represented by a vector of coefficients that linearly reconstruct it from a learned\n dictionary of patches from natural scenes. Next, two measures of saliency (Local and Global) are\n calculated and fused to indicate saliency of each patch. Local saliency is distinctiveness of a patch\n from its surrounding patches. Global saliency is the inverse of a patch’s probability of happening\n over the entire image. The final saliency map is built by normalizing and fusing local and global\n saliency maps of all channels from both color systems. Extensive evaluation over four benchmark\n eye-tracking datasets shows the significant advantage of our approach over 10 state-of-the-art\n saliency models.},\n pages = {1-8},\n month = {Jun},\n year = {2012},\n review = {full/conf},\n type = {bu;mod},\n if = {2012 acceptance rate: 26.2%},\n file = {http://ilab.usc.edu/publications/doc/Borji_Itti12cvpr.pdf}\n}","author_short":["Borji, A.","Itti, L."],"author":["Borji, A.","Itti, L."],"abstract":"We introduce a saliency model based on two key ideas. The first one is considering local and global image patch rarities as two complementary processes. The second one is based on our observation that for different images, one of the RGB and Lab color spaces outperforms the other in saliency detection. We propose a framework that measures patch rarities in each color space and combines them in a final map. For each color channel, first, the input image is partitioned into non-overlapping patches and then each patch is represented by a vector of coefficients that linearly reconstruct it from a learned dictionary of patches from natural scenes. Next, two measures of saliency (Local and Global) are calculated and fused to indicate saliency of each patch. Local saliency is distinctiveness of a patch from its surrounding patches. Global saliency is the inverse of a patch’s probability of happening over the entire image. The final saliency map is built by normalizing and fusing local and global saliency maps of all channels from both color systems. Extensive evaluation over four benchmark eye-tracking datasets shows the significant advantage of our approach over 10 state-of-the-art saliency models."},"bibtype":"inproceedings","biburl":"http://ilab.usc.edu/publications/src/ilab.bib","downloads":0,"search_terms":["exploiting","local","global","patch","rarities","saliency","detection","borji","itti"],"title":"Exploiting Local and Global Patch Rarities for Saliency Detection","year":2012,"dataSources":["wedBDxEpNXNCLZ2sZ"]}