It is important to detect man-made objects in a natural background to reduce false detections in long-wave infrared for safety and security applications. The degree of linear polarization (DoLP) is used frequently to ...
详细信息
It is important to detect man-made objects in a natural background to reduce false detections in long-wave infrared for safety and security applications. The degree of linear polarization (DoLP) is used frequently to solve such problems. DoLP can provide important clues for man-made object signatures. On the other hand, DoLP cannot handle the polarization power because of normalization. First, a novel physics-driven power of linear polarization (PoLP) metric is proposed to find optimal infrared polarization conditions analytically. Second, a data-driven infrared polarization method is presented. Few studies have been conducted in terms of polarimetric optimization at a low level. This paper presents a novel polarimetric information utilization method by applying a two-layered neural network with the inverse contrast radiant intensity (CRI) loss function to find physical meaning. The proposed infrared CRI-based optimal polarimetry (ICOP) could extract the lowlevel contribution of each polarimetric image in discriminating artificialobjects in a natural background. After optimization, the learned weights of the polarimetric images were sine-like, which produced optimal object and background separation. The experimental results for the outdoor scenario validated the optimality of the proposed ICOP in man-made objectdetection in a natural background. Finally, the physics-driven PoLP coincided with the data-driven ICOP in man-made objectdetection.
It is known that overlapping tissues cause highly complex projections in chest radiographs. In addition, artificialobjects, such as catheters, chest tubes and pacemakers can appear on these radiographs. It is importa...
详细信息
It is known that overlapping tissues cause highly complex projections in chest radiographs. In addition, artificialobjects, such as catheters, chest tubes and pacemakers can appear on these radiographs. It is important that the anomaly detection algorithms are not confused by these objects. To achieve this goal, the authors propose an approach to train a convolutional neural network (CNN) to detect chest tubes present on radiographs. To detect the chest tube skeleton as the final output in a better manner, non-uniform rational B-spline curves are used to automatically fit with the CNN output. This is the first study conducted to automatically detect artificialobjects in the lung region of chest radiographs. Other automatic detection schemes work on the mediastinum. The authors evaluated the performance of the model using a pixel-based receiver operating characteristic (ROC) analysis. Each true positive, true negative, false positive and false negative pixel is counted and used for calculating average accuracy, sensitivity and specificity percentages. The results were 99.99% accuracy, 59% sensitivity and 99.99% specificity. Therefore they obtained promising results on the detection of artificialobjects.
暂无评论