|
Infrared thermal imaging technology is a high-tech, which mainly research infrared radiation, transmission, conversion, detection and applications. It plays an important role in areas such as military, medical treatment, detective and so on. At present, it also plays an increasingly important role in post-disaster search, rescue, tracking work after the earthquake, fire, and explosions, mine collapses, landslides, etc.At the site of a disaster, infrared search system can detect the existence of survivors in order to save lives. The goal of Infrared Search System is to target the human body from the recognition of the complexity out of the background. Infrared Imaging Devices used in Post-disaster possess the character of small, portable, robust, etc., however its target recognition completely on relief to complete the eyes of rescuer. Low contrast, fuzzy visual effects, and poor layering characterize infrared image. In particular, there are two circumstances exist in the rescue scene. One is the worst environment, which is smoke, fog, and background noise, then the image through infrared Image is blurred, low resolution, low contrast, low signal-to-noise ratio (SNR), and low quality. The second is our goal to detect is only the local of the body, with other parts of the body that is buried or is blocked; therefore having obtained the human body image is a partial image. The actual situation in the above-mentioned give us identification, search and rescue brought a certain degree of difficulty.For above practical problems, the image registration method with the energy accumulation has been proposed in this paper, by different perspectives on the goals, direction detection, and multi-frame images from the accumulation of energy, improves target contrast and signal to noise ratio, enhances the image of the target recognition. The core idea of Infrared image is to ensure that the objectives of the energy accumulated to the same pixel point through the accurate registration. In this paper, the characteristics of point-based registration method is improved the accuracy of registration. First, through wavelet multi-scale analysis of the characteristics of selected edge points, the use of correlation function is determined the characteristics of points of the same name. Second, by affine transformation to be the baseline image and registration image registration in order to achieve an alignment coefficient, least squares fitting algorithm used to reduce errors, improve accuracy of registration. Finally, after registration of the images in the energy accumulation of different perspectives on the same target to detect the infrared images accumulation, the accumulation of the target energy and suppress background noise, is improved signal to noise ratio and contrast of infrared images.Finally, in this article, the algorithm is simulated by the experiment, has been imitated the disaster scene by different angles of the human body to detect infrared fuzzy images, local images. It has divided into with the experiment into four parts, which are whole-person infrared image, holistic fuzzy infrared image, only to see the infrared image of the head, only to see the infrared image of both hands and feet. At the same time, analysis of their experiments is carried out separately. Experimental evaluation results and related data confirmed that the algorithm could be used to enhance the quality of the infrared images to improve the detection, recognition.
|