Search

In spacecraft attitude control, maintaining an accurate estimate of attitude readings is crucial. Due to aging factors of sensors like gyroscopes, emerged drift or bias from the correct rate values lead to decreased pointing accuracy. This paper proposes unsupervised data-driven approach in order to diagnose in early stage abnormal drifts in spacecraft attitude sensors. Three types of faults are injected in the satellite attitude control simulator. The obtained results are compared with other similar data-driven methods. The comparison shows the superiority of our method in terms of missed alarm rate and incorrect detection rate. In addition, our approach does not require prior knowledge and labels about the attitude sensors faults.

The production of Oil & Gas from underground reservoirs involves chemical and mechanical processes that affect well drilling and operation. Many of these processes may eventually cause a problem with the well, resulting in a decrease in production or in equipment failure. This paper deals with fault prognosis during the practical operation of Oil & Gas wells. This work focus on the remaining useful life prediction of the “Spurious Closure of the Downhole Safety Valve” fault. This paper proposes a scheme based on the use of unsupervised machine learning approach and a drift detection mechanism is employed in order to predict the time to failure, real fault scenarios data are used, the proposed scheme is evaluated using different prognosis performance metrics.

In order to contribute to the security of medical image, we present in this paper a blind and robust watermarking technique that allows the integration of the electronic patient’s record into retinal images. In our work, we present an approach developed to find a compromise between capacity and imperceptibility. The proposed blind watermarking approach is based on the color values difference calculation between pixels to determine the substitution to be made. We proposed in this work ten possible variants of this approach to determine which offers the best compromise between the capacity and the imperceptibility. Following the evaluation of the approach by metrics designed for reliability tests, some variants confirm remarkably good results either by objective or subjective evaluation. The capacity and imperceptibility evaluations reveal very encouraging results. We obtained good results concerning the imperceptibility with a PSNR greater than 69 dB for some variants due to the insignificance of the least significant bits modification. An evaluation of the watermark’s robustness demonstrates the proposed schemas generate reasonably robust watermarked samples against various attacks. A high-quality watermark with a normalized cross-correlation greater than 0.9 for some variants is obtained. However, some variants remain weak in terms of robustness because any simple treatment can modify the LSBs, which makes the extraction impossible.

In order to enhance the security of exchanged medical images in telemedicine, we propose in this paper a blind and robust approach for medical image protection. This approach consists in embedding patient information and image acquisition data in the image. This imperceptible integration must generate the least possible distortion. The watermarked image must present the same clinical reading as the original image. The proposed approach is applied in the frequency domain. For this purpose, four transforms were used: discrete wavelets transform, non-subsampled contourlet transform, non-subsampled shearlet transform and discreet cosine transform. All these transforms was combined with Schur decomposition and the watermark bits were integrated in the upper triangular matrix. To obtain a satisfactory compromise between robustness and imperceptibility, the integration was performed in the medium frequencies of the image. Imperceptibility and robustness experimental results shows that the proposed methods maintain a high quality of watermarked images and are remarkably robust against several conventional attacks.

In order to contribute to the security of medical image, we present in this paper a blind and robust watermarking technique that allows the integration of the electronic patient’s record into a computerized tomography scan. In this approach, a discrete wavelet transform is applied to the image before the integration process, then, a topological reorganization of the coefficients of the LL sub-bands is done by the ZigZag scanning method. The obtained coefficients are then combined to integrate the watermark bits. A hash of the electronic patient record being integrated in the image, the integrity of the watermark can easily be verified. After the evaluation of our approach in terms of invisibility and robustness, the experimental results obtained show that our approach offers excellent imperceptibility (with a PSNR above 70 dB) and very good robustness against several geometric and destructive attacks.