In the non-horizon environment, the laser beam propagation is affected by the scattering medium such as air molecules and dust particles, which causes the scattering phenomenon. Traditional methods are difficult to effectively distinguish scattering effects from color changes at the edges of real objects when processing non view laser images, which reduces recognition accuracy. Therefore, a method for automatic recognition of salient region targets has been proposed. In the Lab color space, a reasonable color change threshold is set to reduce the interference of scattering effects on image quality. By combining the advantages of local and global saliency maps, the salient regions in the image are accurately identified, and diverse features such as color, texture, and shape are extracted. Based on this, a support vector machine (SVM) classifier is trained. After training, the support vector machine (SVM) classifier classifies the feature vectors and accurately assigns them category labels, constructing an efficient set of feature vectors, to identify salient region targets in non view laser images. Experimental results demonstrate that the target boundaries identified by this method are clear and complete in shape. The frame rate of the image processing reaches approximately 65.8 frames per second (FPS), which significantly enhances the efficiency of non-view laser image analysis technology.