Objective Quality Assessment for Color-to-Gray Image Conversion.

Area of Science:

• Computer Vision
• Image Processing
• Perceptual Quality Assessment

Background:

• Color-to-Gray (C2G) image conversion is widely used but lacks standardized objective quality evaluation.
• Subjective quality assessment is reliable but time-consuming and inconvenient.
• Existing objective metrics do not adequately capture perceived quality for C2G conversion.

Purpose of the Study:

• To develop an objective quality model for automatically predicting the perceived quality of C2G converted images.
• To introduce a novel metric, the C2G Structural Similarity (C2G-SSIM) index, for this purpose.
• To demonstrate the utility of C2G-SSIM in practical C2G conversion applications.

Main Methods:

• Proposed the C2G Structural Similarity (C2G-SSIM) index, inspired by structural similarity principles.
• Evaluated luminance, contrast, and structure similarities between original color and converted grayscale images.
• Combined similarity components based on image type for an overall quality score.

Main Results:

• The proposed C2G-SSIM index demonstrated strong agreement with subjective quality rankings.
• C2G-SSIM significantly outperformed existing objective quality metrics in evaluating C2G conversion.
• The metric proved effective in automatic parameter tuning and adaptive fusion of C2G images.

Conclusions:

• C2G-SSIM provides a reliable and objective method for assessing C2G image conversion quality.
• The index offers a valuable tool for advancing C2G algorithm development and application.
• This work establishes a new benchmark for objective quality evaluation in C2G image conversion.