A survey of displacement research based on image recognition

With the rapid development of computer vision and deep learning techniques, significant progress has been made in displacement research based on image recognition. This technique accurately measures the positional changes of target objects at different time points by analyzing image sequences, thus realizing the monitoring and analysis of displacements. This paper reviews the current status of image recognition-based displacement research and focuses on two research directions: target tracking and motion estimation. Target tracking is one of the focuses of this paper, which aims to track changes in the position of a specific target object in an image sequence. Traditional target tracking algorithms rely heavily on hand-designed features and shallow models, such as optical flow and KCF.In recent years, breakthroughs have been made in target tracking algorithms based on deep learning, such as the Siamese network framework and DeepSORT.These algorithms can automatically learn image features and utilize the powerful learning capabilities of deep neural networks to achieve the precise location of target objects. Motion estimation is another important research direction whose goal is to compute the motion field between images in an image sequence, i.e., the motion vector of each pixel. Traditional motion estimation algorithms rely heavily on block matching methods such as optical flow and sparse optical flow. Deep learning-based motion estimation algorithms can automatically learn image features and utilize the powerful learning capabilities of deep neural networks to achieve precise localization of motion fields. Common algorithms include FlowNet and PWC-Net.In addition, this paper will analyze the challenges and future research directions in this field, such as improving robustness, computational efficiency, and multimodal fusion.