Information fusion method of kinect depth camera and thermal infrared camera

Abstract

The invention discloses an information fusion method of a kinect depth camera and a thermal infrared camera, which comprises selecting a calibration board; using a near-infrared short-wave camera and an RGB camera of the kinect depth camera to collect images at different angles on the calibration board; Heat treatment, and use the thermal infrared camera and the RGB camera of the kinect depth camera to obtain pictures of different angles of the calibration plate; use the pinhole imaging model and the principle of dual-target calibration to obtain the external parameters of the near-infrared short-wave camera, RGB camera and thermal infrared camera respectively; Obtain the geometric relationship between the near-infrared short-wave camera and the RGB camera; obtain the geometric relationship between the thermal infrared camera and the RGB camera; according to the geometric relationship between the near-infrared short-wave camera and the RGB camera and the thermal infrared camera and the RGB camera Geometric relationship, information fusion of thermal infrared camera, near infrared shortwave camera and RGB camera.

Description

technical field [0001] The invention relates to the field of structured light multi-sensor information fusion, in particular to an information fusion method of a Kinect depth camera and a thermal infrared camera. Background technique [0002] Compared with traditional depth information acquisition methods, Kinect has the characteristics of portability and low price. Kinect can be used to directly obtain the depth information of indoor scenes and target areas. The combination of RGB information can reconstruct the real 3D information of the scene and the target area. However, based on the limitations of the Kinect imaging principle, Kinect has poor image quality in rainy days, foggy days, and low light conditions; in addition, Kinect’s depth camera and RGB camera have limited calibration parameters at the factory, resulting in unsatisfactory reconstruction results. In traditional machine learning and deep learning, only relying on texture information and three-dimensional in...

AI Technical Summary

Problems solved by technology

However, based on the limitations of the Kinect imaging principle, Kinect has poor image quality in rainy days, foggy days, and low light conditions; in addition, Kinect’s depth camera and RGB camera have limited calibration parameters at the factory, resulting in unsatisfactory reconstruction results. In traditional machine learning and deep learning, only relying on texture information and three-dimensional information is difficult to meet the requirements of input information. Therefore, it is necessary to provide more reliable conditions as input information in the process of machine learning and deep learning to improve the recognition rate of the system.

Images