Light Field Image Quality Assessment Using Edge Similarity

With the development and popularization of virtual reality and augment reality technology in recent years, light field image has been received extensive attentions in academia and industry. However, in the stages of acquisition, compression, storage, transmission and display, light field image are inevitably suffered from a wide various distortions, leading to image quality reduction. Therefore, it is very essential to evaluate the quality of light field image based on the characteristics of human visual system. Considering the structural complexity of light field image, we intend to use edge similarity to construct a mathematical model suitable for the objective assessment of light field image quality. Proposed approach utilizes the gradient and Gabor filter to extract the spatial and frequency edge similarity, and then fuses them together to obtain the edge similarity map, followed by a pooling strategy using frequency edge strength to compute the quality score. Experimental results show that the proposed algorithm is able to better reflect the human perception on light field image, compared with multiple state-of-the-art image quality assessment methods.