Supervised color correction based on QPSO-BP neural network algorithm - Details

Author：

Xu, Xiaozhao (Xu, Xiaozhao.) | Zhang, Xinfeng (Zhang, Xinfeng.) | Cai, Yiheng (Cai, Yiheng.) | Zhuo, Li (Zhuo, Li.) | Shen, Lansun (Shen, Lansun.)

Indexed by：

EI Scopus

Abstract：

Color　information　is　very　important　for　the　applications　of　object　recognition　and　image　retrieval.　However,　the　actual　color　varies　by　the　illumination　conditions.　A　supervised　color　correction　based　on　hybrid　algorithm　combining　Quantum　Particle　Swarm　Optimization　(QPSO)　with　Back　Propagation　(BP)　neural　network　is　proposed　in　this　paper　to　reduce　the　effects　of　illumination　conditions.　Firstly,　the　Macbeth　color　checker　containing　24　color　patches　is　adopted.　Then　those　color　values　of　color　patches　under　unknown　illumination　and　standard　illumination　are　recorded　in　order　to　obtain　the　learning　samples.　Finally,　the　transformation　model　is　established　by　QPSO-BP　neural　network　algorithm　according　to　the　learning　samples.　The　experimental　results　show　that　the　QPSO-BP　algorithm　is　better　than　BP　algorithm　in　convergence　speed.　Comparably,　the　proposed　algorithm　has　better　color　correction　result,　thus　can　be　efficiently　applied　in　practice.　©2009　IEEE.

Keyword：

Neural networks Object recognition Particle swarm optimization (PSO) Color Backpropagation algorithms Image retrieval

Author Community：

[ 1 ] [Xu, Xiaozhao]Signal and Information Processing Lab., Beijing University of Technology, Beijing, China
[ 2 ] [Zhang, Xinfeng]Signal and Information Processing Lab., Beijing University of Technology, Beijing, China
[ 3 ] [Cai, Yiheng]Signal and Information Processing Lab., Beijing University of Technology, Beijing, China
[ 4 ] [Zhuo, Li]Signal and Information Processing Lab., Beijing University of Technology, Beijing, China
[ 5 ] [Shen, Lansun]Signal and Information Processing Lab., Beijing University of Technology, Beijing, China