Daniela Coltuc

Monocular depth estimation and image deblurring are two fundamental tasks in computer vision, given their crucial role in understanding 3D scenes. Performing any of them by relying on a single image is an ill-posed problem. The recent advances in the field of Deep Convolutional Neural Networks (DNNs) have revolutionized many tasks in computer vision, including depth estimation and image deblurring. When it comes to using defocused images, the depth estimation and the recovery of the All-in-Focus (Aif) image become related problems due to defocus physics. Despite this, most of the existing models treat them separately. There are, however, recent models that solve these problems simultaneously by concatenating two networks in a sequence to first estimate the depth or defocus map and then reconstruct the focused image based on it. We propose a DNN that solves the depth estimation and image deblurring in parallel. Our Two-headed Depth Estimation and Deblurring Network (2HDED:NET) extends a conventional Depth from Defocus (DFD) networks with a deblurring branch that shares the same encoder as the depth branch. The proposed method has been successfully tested on two benchmarks, one for indoor and the other for outdoor scenes: NYU-v2 and Make3D. Extensive experiments with 2HDED:NET on these benchmarks have demonstrated superior or close performances to those of the state-of-the-art models for depth estimation and image deblurring.

We propose an innovative delta-differencing algorithm that combines software-updating methods with LZ77 data compression. This software-updating method relates to server-side software that creates binary delta files and to client-side software that performs software-update installations. The proposed algorithm creates binary-differencing streams already compressed from an initial phase. We present a software-updating method suitable for OTA software updates and the method’s basic strategies to achieve a better performance in terms of speed, compression ratio or a combination of both. A comparison with publicly available solutions is provided. Our test results show our method, Keops, can outperform an LZMA (Lempel–Ziv–Markov chain-algorithm) based binary differencing solution in terms of compression ratio in two cases by more than 3% while being two to five times faster in decompression. We also prove experimentally that the difference between Keops and other competing delta-creator software increases when larger history buffers are used. In one case, we achieve a three times better performance for a delta rate compared to other competing delta rates.

Monocular depth estimation is a fundamental challenge since the foundation of computer vision with many real-world applications. Recently, the introduction of Deep Convolutional Neural Networks (CNN) has brought significant improvements to this particular problem. There are many solutions for scene depth estimation with a focus on obtaining high-quality depth maps from a given RGB image. The insertion of prior information by adding a smoothing regularization has improved the results. However, the smoothing of the surfaces comes together with a certain degradation of the edges. The goal of this paper is to make a comparison between various regularization terms used either in supervised or self-supervised learning methods. In addition to this, we modified the regularization term used currently in self-supervised methods such to work in a supervised manner. The experimental results on NYU-Depth v2 have shown that the regularization based on L1 norm of the gradient is the best and the self-supervised modified one outperforms the rest. Finally, rather than relying on common evaluation metrics, we used an additional accuracy measure based on the Steerable Pyramid and Kullback-Leibler divergence (KLD) for edge accuracy of estimated depths that is more sensitive to positional errors of the edges.

One of the challenges of future space missions is the deployment of instruments distributed across multiple satellites that fly in an autonomous formation. This type of mission requires that the spacecraft communicate using inter-satellite links (ISL) and collaborate to fulfil the role of a large-scale, complex scientific instrument. In this context, precise information regarding the relative positioning of the satellites in terms of distance and orientation is essential. In radio frequency (RF) based techniques, obtaining the relative orientation between the satellites starts with estimating the angle-of-arrival (AoA) of the radio signal. This paper proposes an AoA estimation technique that can be used in Orthogonal Frequency-Division Multiplexing (OFDM) communications. Our approach consists in obtaining the AoA using the ISL, specifically the preamble whose primary roles are data link synchronization and channel estimation. Our technique allows a more accurate measurement of the AoA due to an improved ambiguity resolution of phase measurements. We give theoretical results regarding the accuracy of the AoA estimator and we validate them by tests on a real-time functioning testbed consisting in multiple National Instruments Universal Software Radio Peripheral platforms. We show that the carrier frequency offset introduces only a negligible bias in the AoA estimation and we discuss the effect of multipath propagation.

Single Pixel Camera (SPC) is an alternative architecture for imaging systems that emerged as an application of Compressive Sensing theory. The SPC distinctive feature is the single detector that measures linear combinations of pixels instead of pixels as in imaging array cameras. The measurements are obtained by projecting the scene on a Spatial Light Modulator (SLM), the other distinctive component of the SPC architecture. This paper discusses the problem of resolution in SPC. We show that in terms of Point Spread Function (PSF), the resolution model is similar to that of an imaging array camera except the fact that the role of the imaging array is taken over by the SLM. This reduces the contribution of the optics to the SPC resolution comparing with the CCD cameras. Relying on an existing method, we develop a procedure that uses a circular chart to measure the camera PSF. Our procedure has as additional facilities the correction of chart tilt and the capability to work with non-circular apertures. At low noise, tilts up to 15 degrees are practically removed.

The increasing amount of data generated by space applications poses several challenges due to limited resources available onboard: power, memory, computation, data rate. In this paper, we propose Compressed Sensing (CS) as the key tool to face those challenges via compressive imaging. This signal processing technique, only recently applied to space applications, dramatically simplifies the image acquisition featuring native compression/encryption and enabling onboard image analysis, allowing to design simpler and lighter optical systems. In this paper, we try to answer the following question: To what extent are the potential benefits of CS going to materialize in a realistic "space big data" application scenario? To this purpose, we first review compressive imaging techniques and already existing prototypes and concepts, critically discussing the technological issues involved. Then, we propose a set of instrument concepts in the application domains of space science, planetary exploration and earth observation, most suitable for a CS-based application. For the most promising of them, we go deeper into the analysis showing preliminary reconstruction performance tests.

This paper investigates the usefulness of the normalized compression distance (NCD) for image similarity detection. Instead of the direct NCD between images, the paper considers the correlation between NCD based feature vectors extracted for each image. The vectors are derived by computing the NCD between the original image and sequences of translated (rotated) versions. Feature vectors for simple transforms (circular translations on horizontal, vertical, diagonal directions and rotations around image center) and several standard compressors are generated and tested in a very simple experiment of similarity detection between the original image and two filtered versions (median and moving average). The promising vector configurations (geometric transform, lossless compressor) are further tested for similarity detection on the 24 images of the Kodak set subject to some common image processing. While the direct computation of NCD fails to detect image similarity even in the case of simple median and moving average filtering in 3 × 3 windows, for certain transforms and compressors, the proposed approach appears to provide robustness at similarity detection against smoothing, lossy compression, contrast enhancement, noise addition and some robustness against geometrical transforms (scaling, cropping and rotation).