Low-illumination image target detection method based on image fusion and target detection network

Abstract

The invention relates to a low-illumination image target detection method based on image fusion and a target detection network, and the method combines a non-subsampled shearlet transform image fusionalgorithm and an improved target detection network YOLOV4, and comprises the steps: 1) employing an infrared and low-illumination visible light image fusion technology, so that the information complementation can be performed on infrared and low-illumination visible light images, and texture information is highlighted while the contour definition is improved; 2) sending the fused image to an improved YOLOV4 target detection network for detection, and outputting target information under a low-illumination condition; and 3) in order to improve the feature extraction capability of the YOLOV4 network, replacing the residual blocks in the YOLOV4 backbone network with dense link blocks, wherein compared with the residual blocks, the dense link blocks enable the network to improve the feature expression capability and improve the feature extraction capability of the network; finally, experiments prove that the method can improve the target detection capability under the condition of low illumination.

Description

technical field [0001] The invention belongs to the field of machine vision and image processing, and relates to a low-illuminance image target detection method based on image fusion and target detection network. Background technique [0002] Target detection and recognition refers to the use of certain technical means to locate and identify the target of interest in the image. There are currently two main types of object detection methods, sliding window-based methods and region-based objectness-based methods. The former uses the window to traverse the entire image, extracts the image features in the window at each position, and finally uses the classifier to classify the features; the latter uses the method of saliency detection or targetness to extract the candidate area of ​​​​the target, avoiding the sliding window A large number of calculations, but due to the possibility of framing part of the target or poorly distinguishing adjacent targets, it has a certain impact ...

AI Technical Summary

Problems solved by technology

The former uses the window to traverse the entire image, extracts the image features in the window at each position, and finally uses the classifier to classify the features; the latter uses the method of saliency detection or targetness to extract the candidate area of ​​​​the target, avoiding the sliding window A large number of calculations, but due to the possibility of framing part of the target or poorly distinguishing adjacent targets, it has a certain impact on the detection results. Many methods have been proposed to improve this problem. For example, some literature proposes a more targeted area. Extraction method, which reduces the number of candidate windows and improves the quality of candidate windows

[0003] Although the current deep learning network has been greatly developed in the field of target detection, it is basically based on visible light images for target recognition. When the ambient light conditions are poor, the accuracy will drop sharply.

Although infrared images can make up for the impact of poor lighting conditions to a certain extent, it is difficult to accurately identify targets using infrared images alone for target recognition tasks due to the lack of texture and detail information in infrared images.

Images