Automatic archaeological graph generation method based on generative adversarial network

Abstract

The invention discloses an automatic archaeological graph generation method based on a generative adversarial network. The method comprises the following steps: constructing a training data set according to a corresponding relation between an existing cultural relic photo and a standard line graph; utilizing the training data set to train a generative adversarial network, wherein the generative adversarial network comprises a generator and a judging device, the generator is used for generating a predicted line graph according to the cultural relic photo, the judging device is used for restraining the loss between the predicted line graph output by the generator and a standard line graph to meet a set target, and then an optimized generator is obtained. The generative adversarial network is introduced into the field of archaeological graphs for the first time, and the labor cost of archaeological graph drawing is reduced. In addition, the interaction system can expand the data set in the use process, and the problem that the archaeological graph data set is difficult to manufacture is solved.

Description

technical field [0001] The invention relates to the technical field of image processing, and more specifically, to a method for automatically generating archaeological line maps based on a generative confrontation network. Background technique [0002] Deep learning is an extremely important branch of machine learning. With the wave of informatization, computer computing speed and data volume continue to increase, and deep learning technology based on convolutional neural network (CNN) has developed rapidly. For example, successful applications in the commercial field have been achieved in face recognition and language translation. Compared with the traditional discriminative deep learning network framework, the generative deep learning network framework can better explore the essential characteristics of the data, but there are problems such as training difficulties and the need for prior knowledge. [0003] As an important means in archaeological research, archaeological ...

AI Technical Summary

Problems solved by technology

However, traditional edge detection networks such as HED, CED, and BDCN have high requirements for data sets when the outline of supervised data lines is complex, and are prone to problems such as training non-convergence.

In addition, for archaeological line map data sets, the production cost is high, the production cycle is long, and manual drawing is usually required

This leads to difficulties in making data sets, and it is difficult to increase the number of samples in a short period of time.

Images