Marco Diani

Generalized likelihood ratio test approach to multiframe infrared detection of moving objects

This paper investigates the problem of detecting airborne targets in a sequence of images recorded by a long range InfraRed (IR) sensor. The target appears in the IR images as a small, weak signal embedded in a strong background clutter. It is assumed that the target’s amplitude, velocity and position are unknown parameters. To accommodate the unknown parameters the Generalized Likelihood Ratio Test (GLRT) detector is derived. The detector structure and its actual implementation are discussed in detail. To test the detection algorithm an experiment involving a cooperating aircraft has been performed. The preliminary results obtained on this set of experimental data are presented and discussed.

New background subspace basis selection criterion for clutter cnacellation in infrared naval surveillance systems

A clutter removal procedure for Infra-Red (IR) naval surveillance systems is presented. The proposed method is specifically designed to manage the maritime scenario and it is not sensitive to the sharp transition between sea and sky across the horizon line. It is also effective for the removal of striping noise which arises as a consequence of the non-uniform calibration of the detector array. The effectiveness of the clutter removal procedure is illustrated on a set of experimental IR data.

Target detection experiments with a non-parametric detector on a new hyperspectral data set

Target detection experiments with a novel non-parametric detector are carried out exploiting the availability of a new hyperspectral data set featuring a suburban scene with several different targets. Benefiting from its non-parametric nature and from its data adaptivity deriving from the variable-bandwidth approach, the detector is shown to provide promising results for the detection of the targets of interest both in global and local configurations.

A theoretical Gaussian framework for anomalous change detection in hyperspectral images

Exploitation of temporal series of hyperspectral images is a relatively new discipline that has a wide variety of possible applications in fields like remote sensing, area surveillance, defense and security, search and rescue and so on. In this work, we discuss how images taken at two different times can be processed to detect changes caused by insertion, deletion or displacement of small objects in the monitored scene. This problem is known in the literature as anomalous change detection (ACD) and it can be viewed as the extension, to the multitemporal case, of the well-known anomaly detection problem in a single image. In fact, in both cases, the hyperspectral images are processed blindly in an unsupervised manner and without a-priori knowledge about the target spectrum. We introduce the ACD problem using an approach based on the statistical decision theory and we derive a common framework including different ACD approaches. Particularly, we clearly define the observation space, the data statistical distribution conditioned to the two competing hypotheses and the procedure followed to come with the solution. The proposed overview places emphasis on techniques based on the multivariate Gaussian model that allows a formal presentation of the ACD problem and the rigorous derivation of the possible solutions in a way that is both mathematically more tractable and easier to interpret. We also discuss practical problems related to the application of the detectors in the real world and present affordable solutions. Namely, we describe the ACD processing chain including the strategies that are commonly adopted to compensate pervasive radiometric changes, caused by the different illumination/atmospheric conditions, and to mitigate the residual geometric image co-registration errors. Results obtained on real freely available data are discussed in order to test and compare the methods within the proposed general framework.

Real-time moving objects detection and tracking from airborne infrared camera

Detecting and tracking moving objects in real-time from an airborne infrared (IR) camera offers interesting possibilities in video surveillance, remote sensing and computer vision applications, such as monitoring large areas simultaneously, quickly changing the point of view on the scene and pursuing objects of interest. To fully exploit such a potential, versatile solutions are needed, but, in the literature, the majority of them works only under specific conditions about the considered scenario, the characteristics of the moving objects or the aircraft movements. In order to overcome these limitations, we propose a novel approach to the problem, based on the use of a cheap inertial navigation system (INS), mounted on the aircraft. To exploit jointly the information contained in the acquired video sequence and the data provided by the INS, a specific detection and tracking algorithm has been developed. It consists of three main stages performed iteratively on each acquired frame. The detection stage, in which a coarse detection map is computed, using a local statistic both fast to calculate and robust to noise and self-deletion of the targeted objects. The registration stage, in which the position of the detected objects is coherently reported on a common reference frame, by exploiting the INS data. The tracking stage, in which the steady objects are rejected, the moving objects are tracked, and an estimation of their future position is computed, to be used in the subsequent iteration. The algorithm has been tested on a large dataset of simulated IR video sequences, recreating different environments and different movements of the aircraft. Promising results have been obtained, both in terms of detection and false alarm rate, and in terms of accuracy in the estimation of position and velocity of the objects. In addition, for each frame, the detection and tracking map has been generated by the algorithm, before the acquisition of the subsequent frame, proving its capability to work in real-time.

Real-time detection of small and dim moving objects in IR video sequences using a robust background estimator and a noise-adaptive double thresholding

We developed an algorithm for automatically detecting small and poorly contrasted (dim) moving objects in real-time, within video sequences acquired through a steady infrared camera. The algorithm is suitable for different situations since it is independent of the background characteristics and of changes in illumination. Unlike other solutions, small objects of any size (up to single-pixel), either hotter or colder than the background, can be successfully detected. The algorithm is based on accurately estimating the background at the pixel level and then rejecting it. A novel approach permits background estimation to be robust to changes in the scene illumination and to noise, and not to be biased by the transit of moving objects. Care was taken in avoiding computationally costly procedures, in order to ensure the real-time performance even using low-cost hardware. The algorithm was tested on a dataset of 12 video sequences acquired in different conditions, providing promising results in terms of detection rate and false alarm rate, independently of background and objects characteristics. In addition, the detection map was produced frame by frame in real-time, using cheap commercial hardware. The algorithm is particularly suitable for applications in the fields of video-surveillance and computer vision. Its reliability and speed permit it to be used also in critical situations, like in search and rescue, defence and disaster monitoring.