Panoramic Image Quality Evaluation Method Based on Multi-Feature Fusion

Panoramic images are different from traditional 2D images, panoramic images can provide viewers with a broader perspective. Based on the human visual system and panoramic images, this paper proposes a new quality evaluation method for panoramic images based on calculating the feature. human visual system and panoramic images, this paper proposes a new quality evaluation method for panoramic images based on calculating feature-similarity in CMP format. The method firstly converts the ERP format image to CMP format image, i.e., the features under CMP format for similarity calculation. Then the structural and texture feature similarity of the image is calculated using the second-order phase consistency mutual information and local entropy similarity. Then the structural and texture feature similarity of the image is calculated using the second-order phase consistency mutual information and local entropy similarity. In addition, the method also extracts the gradient features and color features of the reference image and the distorted image, and calculates the similarity between them respectively. Finally, these four feature similarities are fused to obtain the objective quality score of the panoramic image. Experimental results on the OIQA dataset show that the method proposed in this paper has better performance than the state-of-the-art OIOA method. state-of-the-art OIOA model.