Infrared small target intelligent identification method based on progressive hierarchical fusion network

Abstract

The invention relates to an infrared small target intelligent identification method based on a progressive hierarchical fusion network, and the method comprises the steps of carrying out the data fusion enhancement from a data dimension through the complementarity of dual-band infrared in imaging information, and improving the available information amount of a neural network for infrared small target identification; building a dual-band infrared data fusion and target detection integrated neural network from embedded application deployment, using weight sharing of parameters among different components in the network fully, enhancing the coupling effect among data, and reducing the overall calculation complexity of a network model. On the basis of dual-band infrared data fusion, a pixel level-feature level-decision level progressive multi-level fusion integrated neural network architecture oriented to infrared weak and small target detection and recognition is designed, efficient infrared small target recognition capability is achieved, and the invention plays an important role in the aspects of night vision monitoring, dangerous source detection, personnel search and rescue and the like.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to an intelligent identification method for infrared small targets based on a progressive hierarchical fusion network. Background technique [0002] Infrared small target detection in complex backgrounds is one of the current research problems in the field of infrared image processing. Since the highlighted background areas and ordinary background edges in the original image will seriously interfere with target detection, high detection rates can be achieved in complex backgrounds. , low false alarm rate, high real-time detection becomes very difficult. [0003] In recent years, with the support of big data, intelligent target recognition technology can extract high-level feature information of targets from multiple dimensions through target feature learning, which has become an important technical approach to improve the accuracy and universality of target recognition. How...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is that there are deficiencies in the recognition of small targets based on single-band infrared images, and the intelligent detection and recognition algorithm for small targets based on multi-band infrared fusion data often combines data fusion with the target when performing fusion operations in the data dimension. The detection and independent design of the network is operated separately, resulting in separation of network structure, high computational complexity, and difficulty in end-to-side deployment; the intelligent recognition method for small infrared targets based on progressive hierarchical fusion networks provided by the present invention makes full use of dual-band infrared information to improve infrared The ability to identify small targets reduces the overall computational complexity of the network and facilitates the deployment of end-to-end applications

Images