Infrared small target detection method based on attention mechanism

Abstract

The invention belongs to the technical field of image processing and computer vision, and particularly relates to an attention mechanism-based infrared small target detection method, which comprises the following steps of: respectively carrying out channel stacking on a Gaussian noise image and an all-zero image with an original image; the channel stack image is transmitted to an attention mechanism module for background suppression and target enhancement; the output image of the attention mechanism module and the original image are subjected to channel stacking and then transmitted to an infrared small target detection module, and a detection image of the infrared small target is obtained; reversely optimizing network parameters of the infrared small target detection module by taking the difference between the detection image and the label image as a loss function; if the loss function is reduced to an acceptable threshold range, training is completed; the to-be-detected image and the all-zero image are subjected to channel stacking to serve as a test sample to be transmitted to the attention mechanism module, an output image of the attention mechanism module and an original image are subjected to channel stacking, and a detection task is achieved through the infrared small target detection module; according to the invention, background clutters can be effectively suppressed, and small target detection is realized.

Description

technical field [0001] The invention belongs to the technical field of image processing and computer vision, and in particular relates to an infrared small target detection method based on an attention mechanism. Background technique [0002] Infrared small target detection plays an important role in the research of infrared imaging systems, and is widely used in tasks such as early warning and long-distance target tracking. However, when cloud clutter or sea clutter is present, small objects are usually submerged in the complex background with low signal-to-noise ratio. In addition, due to the long imaging distance, small objects do not have specific shapes and textures. Therefore, infrared small target detection has been considered to be a difficult and challenging task. [0003] At present, there are mainly four types of infrared small target detection methods: the first is the infrared small target detection method based on background features, which can be divided int...

AI Technical Summary

Problems solved by technology

However, due to the influence of background clutter, there are usually a large number of false alarms in the detection results.

At the same time, the background clutter information of infrared small target images has a high degree of repetition, which makes the network prone to over-fitting.

Images