A detection method for fusing visible light images and corresponding night vision infrared images

Abstract

The invention discloses a detection method based on fusion of a visible light image and a corresponding night vision infrared image, and belongs to the field of machine vision. The method comprises the specific steps as follow: for a visible light image, getting a saliency image of the visible light image using a deep learning network and an incremental coding length method; for a night vision infrared image, getting a saliency image of the night vision infrared image using a GBVS model and a spectrum scale space method based on a hyper-complex frequency domain; and fusing the two saliency images to get a final saliency image. The time domain and the frequency domain are fused. In view of the fact that energy radiation difference is also a factor influencing saliency, the night vision infrared image is processed. The saliency region of the visible light image is extracted using the incremental coding length method. The method is not limited to extraction of low-level features, but also can be used to extract high-level semantic features through a deep learning network. Therefore, the extraction of the saliency region is more comprehensive and more accurate.

Description

technical field [0001] The invention belongs to the field of machine vision, and in particular relates to a detection method for fusing visible light images and corresponding night vision infrared images. Background technique [0002] The machine vision saliency mechanism can be used to extract the most important area of ​​saliency in the road scene, thereby reducing the amount of calculation for later processing. There is also a visual attention mechanism in the human visual system, which can help humans in a short time Concentrate on the most significant and important area to make the fastest response. Inspired by this, researchers also hope that machine vision can also combine complex natural images like human visual attention mechanism. Content screening and simplification of a simplified data, so a visual saliency model for machine vision is proposed. [0003] The visual attention system is an important pre-processing system of the human visual system. It can efficient...

AI Technical Summary

Problems solved by technology

[0008] Aiming at the problems faced by existing visual attention models in target detection in complex road scenes at night, the present invention proposes a detection method that fuses visible light images and corresponding night vision infrared images

Images