Abstract:Traditionally, inertial navigation sensor technology, based on signal sensing technology, has been used to locate search-and-rescue personnel by combining the distance they have moved with their angle of direction. This method is easily affected by poor weather conditions and partial occlusion, and the resulting search-and-rescue efficiency and accuracy are low. In this paper, we propose a new search-and-rescue method based on visual imagery for victims in earthquake areas. Initial visual images of the seismic area are collected by image acquisition equipment, and the wavelet denoising method is used to reduce the noise and improve the discernibility of the images. Then, after noise reduction, the visual image features of people in the seismic area are extracted using the color contrast method, and compared with the original image features of the seismic area, to obtain candidate images of rescuers in the seismic area. According to these candidate images, a Kalman-filter tracking algorithm and a mean-shift tracking algorithm are used to track search-and-rescue personnel in complicated seismic areas during poor weather conditions and partial occlusion, respectively. The experimental results show that the recall rate of the proposed method is above 98.5%, the average accuracy is about 98%, and the average search-and-rescue time is about 23 s. These results indicate that the proposed method can efficiently and accurately locate victims for search-and-rescue operations in seismic areas.