Using a Novel Image Analysis Metric to Calculate Similarity of Input Image and Images Generated by WAE
Vishesh Devgan, Vibhor Singh, Ankit Jain, Ishu Anand, Narina Thakur · 2019
Image similarity is a core concept in Image analysis due to its extensive application in computer vision. Our study of Quasi Euclidean distance metric displayed a potential in the metric viz-a-viz image similarity computation. Here, it has been used to calculate similarity between input images and images generated by the generative model, Wasserstein Auto-Encoder. Standard GANs are great generative models, but has the concerned training instability. Therefore, we have used Wasserstein Auto-Encoder, which also has proven to generate more accurate images, to produce images on which the performance of the metric has been observed.