Visual significance prediction method based on generative adversarial network

Abstract

The invention provides a visual significance prediction method based on a generative adversarial network. which mainly comprises the steps of constructing a saliency generative adversarial network (SalGAN) based on two deep convolutional neural network (DCNN) modules, and training. Specifically, the method comprises the steps of constructing the saliency generative adversarial network (SalGAN) based on two deep convolutional neural network (DCNN) modules, wherein the saliency generative adversarial network comprises a generator and a discriminator, and the generator and the discriminator are combined for predicating a visual saliency graph of a preset input image; and training filter weight in the SalGAN through perception loss which is caused by combining content loss and adversarial loss. A loss function in the method is combination of error from the discriminator and cross entropy relative to calibrated true data, thereby improving stability and convergence rate in adversarial training. Compared with further training of individual cross entropy, adversarial training improves performance so that higher speed and higher efficiency are realized.

Description

Technical field [0001] The present invention relates to the field of image analysis, in particular to a visual saliency prediction method based on a generative confrontation network. Background technique [0002] The process of processing and processing of external information by the human visual perception system is selective screening, rather than receiving all. This process is visual saliency. The rapid development of information technology has made the development of imaging equipment technology more and more rapid, and the image data that people are exposed to every day has increased. Therefore, the research on visual saliency prediction methods is particularly important. At present, its more mature applications mainly include: scene rendering, Image compression, active vision and image quality evaluation. However, the traditional image analysis technology is to perform global processing for the overall image information, and the processing of image information has no prior...

AI Technical Summary

Problems solved by technology

[0004] For the problem of low stability, the object of the present invention is to provide a visual saliency prediction method based on generation confrontation network, based on two deep convolutional neural network (DCNN) modules Salience generation confrontation network (SalGAN) architecture, including A generator and a discriminator whose combination aims to predict a visual saliency map given an input image; filter weights in SalGAN are trained by perceptual losses resulting from combining content loss and adversarial loss

Images