A 3-D Log-Gabor Feature Similarity Index for Quality Assessment of Pansharpened Images

<title>Abstract</title> <p>Assessing the quality of pansharpened image remains a challenging topic, due to unavailability of a reference image. Most existing image quality assessment (IQA) metrics evaluate the spectral and spatial quality of pansharpened images separately and often ignore the inter band relationships in the multispectral data. Meanwhile, metrics aligned with the human visual system (HVS) have gained increasing importance, as they focus on perceptually relevant image features. In this paper, a new method for pansharpened image quality assessment is proposed as an extension of the feature similarity (FSIM) index, which has shown strong performance in assessing the quality of grayscale and RGB images by incorporating low-level features that are important to human perception. To adapt FSIM for multispectral imagery, the metric is extended to operate within a three-dimensional framework, enabling simultaneous analysis of spatial and spectral information. This extension is achieved through the use of 3-D log-Gabor filters to extract structural features across spectral bands. Experimental results demonstrate that the proposed metric provides reliable and perceptually consistent quality assessment and outperforms several existing methods in evaluating pansharpened image quality.</p>