Screen content image quality assessment using distortion-based directional edge and gradient similarity maps

Highlights：

• Proposes an efficient approach to determine whether the given screen content image is under a specific distortion (i.e., contrast change and color saturation change).

• Effectively chooses the right color or chrominance channel to process the content of the image, based on the distortion type.

• Employs different fast-computing edge and gradient features to capture content and structure of the image which result in producing edge similarity maps.

• Seamlessly, uses a weighting pooling strategy to combines the obtained maps and produces the final quality score.

摘要

•Proposes an efficient approach to determine whether the given screen content image is under a specific distortion (i.e., contrast change and color saturation change).•Effectively chooses the right color or chrominance channel to process the content of the image, based on the distortion type.•Employs different fast-computing edge and gradient features to capture content and structure of the image which result in producing edge similarity maps.•Seamlessly, uses a weighting pooling strategy to combines the obtained maps and produces the final quality score.

论文关键词：Screen content image,Image quality assessment,Difference of Gaussian,Contrast change,Color saturation change,Distortion detection

论文评审过程：Received 5 April 2021, Revised 10 October 2021, Accepted 31 October 2021, Available online 17 November 2021, Version of Record 19 November 2021.