Research On Infrared And Visible Images Fusion Algorithm With Saliency Detection

The goal of image fusion is to extract the effective information of different images in the same scene into a new image to obtain more comprehensive information,which is more conducive to subsequent processing.Infrared and visible image fusion is an important branch of image fusion,infrared image reflects the thermal radiation of the objects,visible image captures visible light information reflected by objects,and the two have strong complementary characteristics,fusion of these two kinds of images is one of the current hot tasks.At present,infrared and visible light image fusion technology is widely used in security,military,automotive and other fields.This paper mainly studies the infrared and visible image fusion technology based on saliency detection,and the main contents are as follows:1.In order to solve the problem that common saliency detection fusion algorithm fusion image have poor spatial sense of image fusion and easy to introduce noise and artifacts,an image fusion algorithm based on entropy rate superpixel segmentation was designed.In order to extract the highlight of the infrared image targets,firstly,the infrared image is adaptively divided into image blocks according to its features by using entropy rate superpixel segmentation,the salient value of the image block is calculated to obtain the gray salient map of the infrared image,and then the salient region mask of the infrared image is obtained through the threshold segmentation and morphological processing optimization,finally by means of poisson image editing will be salient areas of infrared image seamless cloning to visible image.The experimental results show that the proposed algorithm can retain the infrared features completely,and the fusion image edges are clear and the contrast is high,it is beneficial to target detection and recognition.2.In order to solve the problem of edge blur of the common fusion algorithm which combines saliency detection and multi-scale decomposition,an image fusion algorithm based on interval gradient filter is designed.The infrared image and visible image are decomposed into base layer,detail layer and edge layer at four scales by using Gaussian filter and interval gradient filter.For the base layer,the fusion rule based on saliency detection is used.The source image is divided into image blocks by entropy rate superpixel segmentation,and then the weight of the weighted summation of the base layer is constructed according to the salient value of the image blocks.The small scale of edge and detail layer adopts the fusion rule of maximizing absolute value.The large scale of edge and details layer is fused by constructing feature map and mapping them to weighted summation weight map.Finally,the fused components is superimposed to obtain the fused image.Experimental results show that the proposed algorithm can avoid artifacts while preserving edge details,and the obtained fusion image is consistent with human perception,easy to perceive and understand the scene.