Visual Tracking and Identification of Targets in Limited Spatial Resolution

Visual Tracking and Identification of Targets in Limited Spatial Resolution Extending visual sensing capacity in target tracking and identification is very important for Army’s future force. Contemporary techniques assume and depend on good quality images and video, as visual details need to be used for target modeling. In practice, however, as the environments are unconstrained, their performances are largely degraded when the quality of images and video is seriously degraded. The existing solution to this issue is to perform pre-processing to enhance and restore the quality of the imagery, e.g., via image super-resolution or deblurring, before performing target detection, tracking and recognition. However, such image restoration tasks themselves are very difficult, and they generally are very computationally demanding. Therefore, this solution is not practical. This project is targeted on exploring a general and unified approach that does not perform explicit image restoration as pre-processing in performing target tracking and recognition under lowquality imagery, such as poor image resolution, sever blurs, and other image/video degradation. Although this project is focused on one case study of limited spatial resolution, the proposed new approach is expected to be generally applicable to other situations of visual degradation. Specifically, we plan to address the following issues: � A principled approach and its theoratical foundation. The key idea of avoiding performing explicit image restoration is to embed the prior knowledge for restoration into target matching. Our objective is to develop a general approach of learning image similarity and regression that applies to various situations of low-quality imagery. � Target matching and tracking. The performance of target tracking is largely determined by the metric for image similarity. It should be adaptive to different imagery. Our objective is to learn the metric for limit-resolution imagery by steering the known similarity metric in good-resolution images, based on an exemplar database conveying the knowledge. � Target identification. Targets are seen from very different views. It is very difficult to determine if the low-quality images from different views are the same target. Our objective is to learn the visual similarity metrics to generate the predicted novel views. The difference between the actual image and predicted view will be used for better target identification. One of its innovations is that it avoids performing explicit and dedicated image/video restoration which is in general computationally demanding, but rather using the low-quality data directly with the implicit knowledge learned from data. This research has not been studied before, and it leads to very efficient and innovative solutions with very wide applications. Moreover, it is a principled and general solution that will be able to handle various image degradations in the same framework. Once we have collected the training data (i.e., an exemplar database) that associate the low-quality data with their corresponding high-quality data, the implicit knowledge conveyed in this exemplar database facilitate to learn the similarity metric of the lowquality data, and it will empower effective reconstruction-based local regression. This will make possible target matching, tracking and recognition on low-quality data. In addition, image restoration and enhancement (e.g., super resolution) can also be done in the proposed new approach via reconstruction-based local regression