A New Method for Training Set Construction of Underwater Fish Image Enhancement Applied to Pix2Pix
This paper proposes a method to compare the image enhancement quantization effect of multi scale retina enhancement algorithm with color restoration(MSRCR)method under different parameters by using UIQM(underater image quality measurement)and selects the best one to form a training set,which is employed to train the Pix2Pix model.Experimental results show that high-quality Pix2Pix model parameters for underwater fish image enhancement can be obtained by the training set established through this method,and the method can also be extended to other generative adversarial network(GAN)model training.
underwater fish image enhancementtraining setMSRCRPix2Pix
万方数据
维普
