Contrast enhancement by automatic and parameter-free piecewise linear transformation for color images

Chun Ming Tsai*, Zong Mu Yeh

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

76 Citations (Scopus)


Conventional contrast enhancement methods have four shortcomings. First, most of them need transformation functions and parameters which are specified manually. Second, most of them are application-oriented methods. Third, most of them are performed on gray level images. Fourth, the histogram equalization (HE) based enhancement methods use non-linear transform function. Thus, this paper proposes an automatic and parameter-free contrast enhancement algorithm for color images. This method includes following steps: First, RGB color space is transformed to HSV color space. Second, image content analysis is used to analyze the image illumination distribution. Third, the original image is enhanced by piecewise linear based enhancement method. Finally, the enhancement image is transformed back to RGB color space. This novel enhancement is automatic and parameter-free. Our experiments included various color images with low and high contrast. Experiment results show that the performance of the proposed method is better than histogram equalization (HE) and its six variations in non-over enhancement and natural clearly revealed. Moreover, the proposed algorithm can be run on an embedded environment (such as mobile device, digital camera, or other consumer products) and processed in real-time system due to its simplicity and efficiently.

Original languageEnglish
Pages (from-to)213-219
Number of pages7
JournalIEEE Transactions on Consumer Electronics
Issue number2
Publication statusPublished - 2008 May


  • Brightness
  • Color
  • Helium
  • Histograms
  • Image color analysis
  • Image segmentation
  • Smoothing methods

ASJC Scopus subject areas

  • Media Technology
  • Electrical and Electronic Engineering


Dive into the research topics of 'Contrast enhancement by automatic and parameter-free piecewise linear transformation for color images'. Together they form a unique fingerprint.

Cite this