A generative adversarial network image completion method from a space-based perspective

Abstract

The invention provides a generation confrontation network image completion method from a space-based perspective, which belongs to the technical field of aerial surveillance. In view of the situation that the image captured by the intelligent UAV may be partially missing due to constraints such as communication or weather, the method of the present invention first uses the saliency detection model DeepCut to detect the shape of the object in the image, and then uses Canny edge detection to obtain its outline. Figure, the contour map is incomplete, and then use the generative adversarial network to complete the contour map of the object, and finally input the completed contour map as additional information together with the missing image into the image completion network to obtain the final complete image . The method of the invention can effectively improve the authenticity of the completion, has a remarkable effect, and has great significance in applications such as security detection, intelligence collection and analysis under space-based surveillance, and the like.

Description

technical field [0001] The invention belongs to the technical field of aerial surveillance and relates to a technology for complementing images taken by unmanned aerial vehicles, in particular to a method for complementing images generated against a network from a space-based perspective. Background technique [0002] With the continuous maturity and development of UAV technology, its application fields are becoming more and more extensive. In terms of military affairs, it can perfectly replace manned machines to perform 4D tasks, such as boring, harsh environment, high risk, and deep enemy tasks. For civil use, it is widely used in various fields such as agriculture, industry, and service industries, such as pesticide spraying, fire rescue, geological exploration, archaeological exploration, logistics and transportation, etc. However, when UAVs work under relatively harsh conditions, such as limited communication, bad weather, etc., there are many interferences, it will of...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the current image completion of UAV transmission, which cannot maintain the local consistency of the image, it is difficult to complete the missing of too large size, etc., the present invention proposes a generative adversarial network image completion method from a space-based perspective. Used in applications such as security detection, intelligence collection and analysis under space-based surveillance

Images