Large-scale image colorization based on divide-and-conquer support vector machines

Bo Wei Chen, Xinyu He, Wen Ji, Seungmin Rho, Sun Yuan Kung

Research output: Contribution to journalArticlepeer-review

5 Scopus citations


This study presents a system that can automatically colorize grayscale images in large quantities. To enable big data training, divide-and-conquer support vector machines (SVMs) also proposed as classifiers are frequently used in this study. The system is composed of two components—image classification and local-descriptor classification. The former firstly analyzes an input by using a classifier, so that the system can determine which class should serve as the knowledge base. After the class is decided, the latter stage subsequently uses this knowledge base as the reference to colorize the input. Experimental results showed that the accuracy of classification in image classification could reach 90.50 %. Moreover, in the local-descriptor classification, the majority of pixels were successfully assigned correct colors. During the efficiency test, the proposed divide-and-conquer SVM enhanced computational speed while maintaining the accuracy. Such findings demonstrate the effectiveness of the proposed method and the feasibility of our idea.

Original languageEnglish (US)
Pages (from-to)2942-2961
Number of pages20
JournalJournal of Supercomputing
Issue number8
StatePublished - Aug 1 2016

All Science Journal Classification (ASJC) codes

  • Software
  • Theoretical Computer Science
  • Information Systems
  • Hardware and Architecture


  • Big data
  • Divide-and-conquer support vector machine
  • Image colorization


Dive into the research topics of 'Large-scale image colorization based on divide-and-conquer support vector machines'. Together they form a unique fingerprint.

Cite this