Gwendoline Blanchet

An explicit sharpness index related to global phase coherence

We propose a definition of a Sharpness Index that is closely related to the notion of Global Phase Coherence recently introduced for automatic image restoration and image quality assessment. Using Gaussian random fields instead of random phase images, we can estimate the probability that a random image has a given Total Variation, which leads us to an explicit formula and a fast algorithm. Theoretical arguments and numerical experiments are given to assess the similarity between the Sharpness Index and the Global Phase Coherence, and an application to non-parametric blind deconvolution is presented, that illustrates the possibilities offered by this new approach.

Measuring the Global Phase Coherence of an image

The Fourier phase spectrum of an image is well known to contain crucial information about the image geometry, in particular its contours. In this paper, we show that it is also strongly related to the image quality, in particular its sharpness. We propose a way to define the Global Phase Coherence (GPC) of an image, by comparing the likelihood of the image to the likelihood of all possible images sharing the same Fourier power spectrum. The likelihood is measured with the total variation (Rudin-Osher-Fatemi implicit prior), and the numerical estimation is realized by a Monte-Carlo simulation. We show that the obtained GPC measure decreases with blur, noise, and ringing, and thus provides a new interesting sharpness indicator, that can be used for parametric blind deconvolution, as demonstrated by experiments.

A survey of pansharpening methods with a new band-decoupled variational model

Abstract Most satellites decouple the acquisition of a panchromatic image at high spatial resolution from the acquisition of a multispectral image at lower spatial resolution. Pansharpening is a fusion technique used to increase the spatial resolution of the multispectral data while simultaneously preserving its spectral information. In this paper, we consider pansharpening as an optimization problem minimizing a cost function with a nonlocal regularization term. The energy functional which is to be minimized decouples for each band, thus permitting the application to misregistered spectral components. This requirement is achieved by dropping the, commonly used, assumption that relates the spectral and panchromatic modalities by a linear transformation. Instead, a new constraint that preserves the radiometric ratio between the panchromatic and each spectral component is introduced. An exhaustive performance comparison of the proposed fusion method with several classical and state-of-the-art pansharpening techniques illustrates its superiority in preserving spatial details, reducing color distortions, and avoiding the creation of aliasing artifacts.

Algorithms improvement in image processing for optical observations of artificial objects in geostationary orbit with the TAROT telescopes

TAROT (Télescope á Action Rapide pour les Objets Transitoires - Rapid Action Telescope for Transient Objects) is a network of robotic ground based telescopes. Since 2002, we use them for a survey of artificial objects (satellites, debris) in the geostationary orbit. The objects are detected, their orbit is computed, and follow-up observations are planned. We are currently implementing new, more efficient, image processing algorithms in order to improve the processing speed, the sensitivity, and to decrease the rate of false detections. We present our new implemented methods, as well as the results obtained.

Automatic detection of well sampled images via a new ringing measure

According to Shannon Sampling Theory, Fourier interpolation is the optimal way to reach subpixel accuracy from a properly-sampled digital image. However, for most images this interpolation tends to produce an artifact called ringing, that consists in undesirable oscillations near objects contours. In this work, we propose a way to detect this ringing artifact. Using Euler zigzag numbers, we compute the probability that neighboring gray-levels form an alternating sequence by chance, and characterize these undesirable ringing blocks as structures that would be very unlikely in a random image. We then show two applications where the associated algorithm is used to test or enforce the compliance of an image with Fourier interpolation.

Review of low-baseline stereo algorithms and benchmarks

The purpose of this work is to review and evaluate the performance of several algorithms which have been designed for satellite imagery in a geographic context. In particular we are interested in their performance with low-baseline image pairs like those which will be produced by the Pleiades satellite. In this study local and global state of the art algorithms have been considered and compared: CARMEN, MARC, MARC2 and MICMAC. This paper aims also at proposing a new benchmark to compare stereo algorithms. A set of simulated stereo images for which the ground truth is perfectly known will be presented. The obtained accuracy for the ground truth is more than a hundredth of pixel. The existence of an accurate ground truth is a major improvement for the community, allowing to quantify very precisely the disparity error in a realistic setting.

A linear prefilter for image sampling with ringing artifact control

When sampling a continuous image or subsampling a discrete image, aliasing artifacts can be controlled by filtering the data prior to sampling. Bandlimiting filters completely avoid aliasing artifacts, but have to find a compromise between blur and ringing artifacts. In this paper, we propose a new joint definition of blur/ringing artifacts, that associates to a given bandlimited prefilter its so-called spread-ringing curve. We then build a set of filters yielding the optimal blur/ringing compromise according to the previous definition. We show on experiments that such filters yield sharper images for a given level of ringing artifact.

Discarding moving objects in quasi-simultaneous stereovision

This paper proposes a statistical rejection rule, designed for small baseline stereo satellites. The method learns an a contrario model for image blocks and discards the casual matches between the images of the stereo pair. A formula estimating the expected number of false alarms under the background model is proved. Comparative experiments on quasi-simultaneous stereo in aerial imagery demonstrate the elimination of all incoherent motions.

Fast and accurate denoising method applied to very high resolution optical remote sensing images

Restoration of Very High Resolution (VHR) optical Remote Sensing Image (RSI) is critical and leads to the problem of removing instrumental noise while keeping integrity of relevant information. Improving denoising in an image processing chain implies increasing image quality and improving performance of all following tasks operated by experts (photo-interpretation, cartography, etc.) or by algorithms (land cover mapping, change detection, 3D reconstruction, etc.). In a context of large industrial VHR image production, the selected denoising method should optimized accuracy and robustness with relevant information and saliency conservation, and rapidity due to the huge amount of data acquired and/or archived. Very recent research in image processing leads to a fast and accurate algorithm called Non Local Bayes (NLB) that we propose to adapt and optimize for VHR RSIs. This method is well suited for mass production thanks to its best trade-off between accuracy and computational complexity compared to other state-of-the-art methods. NLB is based on a simple principle: similar structures in an image have similar noise distribution and thus can be denoised with the same noise estimation. In this paper, we describe in details algorithm operations and performances, and analyze parameter sensibilities on various typical real areas observed in VHR RSIs.

BBD: A new Bayesian bi-clustering denoising algorithm for IASI-NG hyperspectral images

We propose a new denoising method for 3D hyperspectral images for the future MetOp-Second Generation series satellite incorporating the new IASI-NG interferometer, to be launched in 2021. This adaptive method retrieves the data model directly from the input noisy granule, using the following techniques: dual clustering (spectral and spatial), dimensionality reduction by adaptive PCA, and Bayesian denoising. The use of dimensionality reduction by PCA has been already proven an effective denoising technique because of intrinsic data redundancy. We demonstrate here that by combining a local PCA dimensionality reduction with a dual clustering and a Bayesian denoising, it is possible to improve significantly the PSNR with respect to PCA reduction alone. This noise reduction hints at the possibility to multiply of the resolution of the satellite by factor 4, while keeping an acceptable SNR.