Image Color-Correction Models based on Digital Photo Content
碩士 === 世新大學 === 資訊管理學研究所(含碩專班) === 98 === Low-end digital camera doesn’t have good auto exposure and white balancing algorithms. It results in over- or under-exposure, pale sky, incorrect skin color and no detail in shadow. To enhance the image quality of those photos, the study proposes three model...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2010
|
Online Access: | http://ndltd.ncl.edu.tw/handle/07474617062472686273 |
id |
ndltd-TW-098SHU05396063 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-098SHU053960632016-05-04T04:31:50Z http://ndltd.ncl.edu.tw/handle/07474617062472686273 Image Color-Correction Models based on Digital Photo Content 基於數位相片內容的影像色彩修正技術 Chi-Neng Huang 黃祺能 碩士 世新大學 資訊管理學研究所(含碩專班) 98 Low-end digital camera doesn’t have good auto exposure and white balancing algorithms. It results in over- or under-exposure, pale sky, incorrect skin color and no detail in shadow. To enhance the image quality of those photos, the study proposes three models as a post image processing to make their color more preferable. It was done by detecting sky, face and grass region first. Then, the ratios between the uncorrected colors and their preferred RGB values were calculated for local color correction. The land was corrected by auto white balancing. The final color correction ratio of sky, face, grass and land were determined by mixing the four according to its vertical location. Normal photos were perturbed randomly as uncorrected photos and their color became more acceptable by using the proposed algorithm. The first model uses weighted gradation map of the four correction ratios in vertical direction to generate smooth color adjustment from top to bottom. The second model uses multi-pass region growing technique to segment sky areas. However, it’s easily affected by buildings and trees in a photo therefore having lower robustness. The third model uses probability density of the sky, skin and grass colors to generate region masks for weighted color adjustment. Psychophysical experiment results show that the third model was preferred by observers. Its mean score is even better than the unperturbed originals. none none 徐道義 孫沛立 2010 學位論文 ; thesis 96 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 世新大學 === 資訊管理學研究所(含碩專班) === 98 === Low-end digital camera doesn’t have good auto exposure and white balancing algorithms. It results in over- or under-exposure, pale sky, incorrect skin color and no detail in shadow. To enhance the image quality of those photos, the study proposes three models as a post image processing to make their color more preferable. It was done by detecting sky, face and grass region first. Then, the ratios between the uncorrected colors and their preferred RGB values were calculated for local color correction. The land was corrected by auto white balancing. The final color correction ratio of sky, face, grass and land were determined by mixing the four according to its vertical location. Normal photos were perturbed randomly as uncorrected photos and their color became more acceptable by using the proposed algorithm.
The first model uses weighted gradation map of the four correction ratios in vertical direction to generate smooth color adjustment from top to bottom. The second model uses multi-pass region growing technique to segment sky areas. However, it’s easily affected by buildings and trees in a photo therefore having lower robustness. The third model uses probability density of the sky, skin and grass colors to generate region masks for weighted color adjustment. Psychophysical experiment results show that the third model was preferred by observers. Its mean score is even better than the unperturbed originals.
|
author2 |
none |
author_facet |
none Chi-Neng Huang 黃祺能 |
author |
Chi-Neng Huang 黃祺能 |
spellingShingle |
Chi-Neng Huang 黃祺能 Image Color-Correction Models based on Digital Photo Content |
author_sort |
Chi-Neng Huang |
title |
Image Color-Correction Models based on Digital Photo Content |
title_short |
Image Color-Correction Models based on Digital Photo Content |
title_full |
Image Color-Correction Models based on Digital Photo Content |
title_fullStr |
Image Color-Correction Models based on Digital Photo Content |
title_full_unstemmed |
Image Color-Correction Models based on Digital Photo Content |
title_sort |
image color-correction models based on digital photo content |
publishDate |
2010 |
url |
http://ndltd.ncl.edu.tw/handle/07474617062472686273 |
work_keys_str_mv |
AT chinenghuang imagecolorcorrectionmodelsbasedondigitalphotocontent AT huángqínéng imagecolorcorrectionmodelsbasedondigitalphotocontent AT chinenghuang jīyúshùwèixiāngpiànnèiróngdeyǐngxiàngsècǎixiūzhèngjìshù AT huángqínéng jīyúshùwèixiāngpiànnèiróngdeyǐngxiàngsècǎixiūzhèngjìshù |
_version_ |
1718260259057827840 |