Real Time Infrared Thermal Signature Detection Using Neural Networks

Thermal imaging is a simple and useful technique for post mission analysis. Combining with health monitoring and fault diagnostic techniques, the method can reduce false alarms and improve the diagnostic capability of naval surface, subsurface, and air assets. It can also be used for detecting corrosion, erosion, and hot spots due to bearing degradation. However, there is a fundamental problem in thermal imaging. There is too much data and it must be processed in real time. There is a technology, often used in other domains, to reduce the amount of data in the image to an alternative set of data which represents the most important features in the image. This technology is called Principal Component Analysis (PCA). Using a dynamic PCA, the amount of data which would need to be searched for the target of interest can be greatly reduced. The reason PCAs have not been used for this application is that dynamically computing the Principal Components from an image was too time consuming. Each component had to be computed independently - first the first component, then the second component, then the third component, etc. Since the computation of each component is itself an iterative process, real-time results could not be achieved. Intelligent Automation, Incorporated (IAI) has solved this problem. With our approach, we will be able to compute the Principal Components of an image in real-time. Our innovation allows us to compute all the Principal Components in parallel, with all values converging at the same rate, so that final values are achieved for all components simultaneously instead of serially. This allows us to perform dynamic PCA in real-time and therefore apply this powerful technology to high speed thermal image processing. Intelligent Automation, Incorporated (IAI) has solved this problem. With our approach, we will be able to compute the Principal Components of an image in real-time. Our innovation allows us to compute all the Principal Components in parallel, with all values converging at the same rate, so that final values are achieved for all components simultaneously instead of serially. This allows us to perform dynamic PCA in real-time and therefore apply this powerful technology to high speed thermal image processing. Intelligent Automation, Incorporated (IAI) has solved this problem. With our approach, we will be able to compute the Principal Components of an image in real-time.