Inter-Layer Prediction of Color in High Dynamic Range Image Scalable Compression

Area of Science:

• Computer Science
• Image Processing
• Video Compression

Background:

• High dynamic range (HDR) video offers superior visual quality but requires efficient compression techniques.
• Scalable video coding allows adaptation to various network conditions and display capabilities.
• Inter-layer prediction (ILP) is crucial for enhancing compression efficiency in scalable video coding.

Purpose of the Study:

• To develop a color inter-layer prediction (ILP) method for scalable coding of HDR video content.
• To investigate the effectiveness of different color representations, specifically Y'CbCr and CIE u'v', for HDR compression.
• To improve upon existing ILP methods by leveraging hue preservation principles.

Main Methods:

• A novel color ILP method was developed for HDR video scalable coding.
• Two color representations (Y'CbCr and CIE u'v') were analyzed for HDR compression.
• Prediction equations were derived based on a color model ensuring hue preservation between HDR and LDR layers.
• The proposed method predicts chromatic components of the HDR layer from the decoded LDR layer.

Main Results:

• The CIE u'v'-based colorspace demonstrated significant advantages for HDR content compression, particularly in scalable contexts.
• The proposed ILP scheme using the CIE u'v' colorspace outperformed the state-of-the-art ILP method.
• Experimental results validated the effectiveness of the hue preservation assumption for accurate color prediction.

Conclusions:

• The developed color ILP method offers improved compression efficiency for HDR video.
• The CIE u'v' colorspace is highly suitable for HDR video compression within a scalable coding framework.
• This research contributes to more efficient and visually appealing HDR video delivery.