Relevance-Based Template Matching for Tracking Targets in FLIR Imagery

Abstract

One of the main challenges in automatic target tracking applications is represented by the need to maintain a low computational footprint, especially when dealing with real-time scenarios and the limited resources of embedded environments. In this context, significant results can be obtained by using forward-looking infrared sensors capable of providing distinctive features for targets of interest. In fact, due to their nature, forward-looking infrared (FLIR) images lend themselves to being used with extremely small footprint techniques based on the extraction of target intensity profiles. This work proposes a method for increasing the computational efficiency of template-based target tracking algorithms. In particular, the speed of the algorithm is improved by using a dynamic threshold that narrows the number of computations, thus reducing both execution time and resources usage. The proposed approach has been tested on several datasets, and it has been compared to several target tracking techniques. Gathered results, both in terms of theoretical analysis and experimental data, showed that the proposed approach is able to achieve the same robustness of reference algorithms by reducing the number of operations needed and the processing time.

A novel method for quantifying target tracking difficulty of the infrared image sequence

“…Part of these sequences is collected from the Army Missile Command (AMCOM) infrared dataset [34,35]; the rest sequences are collected by ourselves. These sequences include both the long-wave infrared and medium-wave infrared sequences.…”

Joint Infrared Target Recognition and Segmentation Using a Shape Manifold-Aware Level Set

“…There are also hybrid techniques [ 54 – 56 ] that combine both CAD models and data-driven 2D image features for model learning and inferencing. Indeed, 2D image features play an important role in many IR ATR algorithms [ 57 ], and a variety of such shape features were evaluated in [ 58 ]. One example of a hybrid technique is the multi-view morphing algorithm that was used in [ 59 ] to construct a view morphing database in an implicit way.…”

Advances in Target Detection and Tracking in Forward-Looking InfraRed (FLIR) Imagery

“…A first way to introduce them could be to consider the heterogeneous application fields tackled, which range from dim target detection from ground-based Earth Observation (EO) or sea vessel-mounted sensors [ 1 , 2 ], to the automatic tracking of humans [ 3 – 6 ], animals [ 7 ] and military vehicle [ 8 , 9 ] targets to semantic annotation of video sequences [ 10 ]. Another interesting point of view that could be adopted relates to the specific issues of detection and tracking in FLIR imagery that are actually addressed by the reported techniques: image registration [ 11 ], target representation and recognition [ 3 , 7 – 9 ], occlusion handling [ 4 , 6 ], multi-target association [ 10 ], clutter removal [ 1 , 2 ] and latency reduction [ 5 ]. Lastly, an interesting perspective is obtained by considering whether the various works propose new algorithms specifically designed to cope with the peculiarities of FLIR imagery [ 2 , 3 , 5 – 7 ], adapt or combine well-known techniques (like sparse coding, particle filtering, etc. )…”

“…Another interesting point of view that could be adopted relates to the specific issues of detection and tracking in FLIR imagery that are actually addressed by the reported techniques: image registration [ 11 ], target representation and recognition [ 3 , 7 – 9 ], occlusion handling [ 4 , 6 ], multi-target association [ 10 ], clutter removal [ 1 , 2 ] and latency reduction [ 5 ]. Lastly, an interesting perspective is obtained by considering whether the various works propose new algorithms specifically designed to cope with the peculiarities of FLIR imagery [ 2 , 3 , 5 – 7 ], adapt or combine well-known techniques (like sparse coding, particle filtering, etc. ) often used with visible light images to suit the requirements of the infrared spectrum [ 1 , 4 , 8 ], or simply study how existing alternatives would perform in the considered domain with the aim to stimulate further research [ 9 , 11 ].…”

“…Paravati and Esposito [ 5 ] consider the problem of ATR in FLIR imagery from the point of view of computational complexity and propose a mechanism for further improving the performance of high-speed tracking methods based on template matching by reducing the domain space using a relevance-based technique. In particular, they focus on a well-known tracking filter with an extremely small kernel based on intensity variation function and use a dynamic threshold on the correlation degree computed by the above function on consecutive frames to reduce the number of points to be processed.…”