Recognition and Depth Estimation of Ships Based on Binocular Stereo Vision

Methods based on light field information have shown promising results in depth estimation and underwater image restoration. However, improvements are still needed in terms of depth estimation accuracy and image restoration quality. Previous work on underwater image restoration employed an image formation model (IFM) that overlooked the effects of light attenuation and scattering coefficients in underwater environments, leading to unavoidable color deviation and distortion in the restored images. Additionally, the high blurriness and associated distortions in underwater images make depth information extraction and estimation very challenging. In this paper, we refine the light propagation model and propose a method to estimate the attenuation and backscattering coefficients of the underwater IFM. We simplify these coefficients into distance-related functions and design a relationship between distance and the darkest channel to estimate the water coefficients, effectively suppressing color deviation and distortion in the restoration results. Furthermore, to increase the accuracy of depth estimation, we propose using blur cues to construct a cost for refocusing in the depth direction, reducing the impact of high signal-to-noise ratio environments on depth information extraction, and effectively enhancing the accuracy and robustness of depth estimation. Finally, experimental comparisons show that our method achieves more accurate depth estimation and image restoration closer to real scenes compared to state-of-the-art methods.

Depth estimation and camera calibration of a focused plenoptic camera for visual odometry

In this paper, we propose MINE to perform novel view synthesis and depth estimation via dense 3D reconstruction from a single image. Our approach is a continuous depth generalization of the Multiplane Images (MPI) by introducing the NEural radiance fields (NeRF). Given a single image as input, MINE predicts a 4-channel image (RGB and volume density) at arbitrary depth values to jointly reconstruct the camera frustum and fill in occluded contents. The reconstructed and inpainted frustum can then be easily rendered into novel RGB or depth views using differentiable rendering. Extensive experiments on RealEstate10K, KITTI and Flowers Light Fields show that our MINE outperforms state-of-the-art by a large margin in novel view synthesis. We also achieve competitive results in depth estimation on iBims-1 and NYU-v2 without annotated depth supervision. Our source code is available at https://github.com/vincentfung13/MINE.

Image stitching refers to the combination of multiple images with a certain over lappingarea rate acquired by the image acquisition device into one image, which has a full picture of each image. Image stitching technology is an important part in the field of digital image processing. With the increasing requirements for image quality and the limited resolution of ordinary cameras, it is impossible to obtain large angle of view and clear image at the same time. In this paper, the key technologies involved in pavement crack image stitching are intensively studied, including image feature point detection and matching, internal and external camera parameter estimation and optimization, and image fusion. The work done in this paper and the main innovations include: (1)Two main feature detection and description algorithms, SIFT algorithm and ORBalgorithm, are studied. In view of the fact that the SIFT algorithm has redundancy when the Gaussian pyramid is established, this paper appropriately reduces the number of groups and maintains the number and stability of feature point detection. Combined with the advantages and disadvantages of the improved SIFT algorithm and ORB algorithm, and the performance of the two algorithms in the detection of pavement cracks, the improvedSIF'T algorithm is used as the feature point detection algorithm, and rBRlEF algorithm for describing the feature points in the ORB algorithm is used as the feature point description algorithm. The description algorithm is called the BSIF'T algorithm. Experiments show that the robustness of BSIFT algorithm is similar to that of SIFT algorithm, which is better than ORB algorithm and faster than SIF'T algorithm. (2)Image fiusion algorithms are studied, including best seam-line search and image fusion. Aiming at the ghost phenomenon that may occur in overlapping regions, this paper proposes a best seam-line search algorithm based on distance transform.

BackgroundNon-invasive planar fluorescence reflectance imaging (FRI) is used for accessing physiological and molecular processes in biological tissue. This method is efficiently used to detect superficial fluorescent inclusions. FRI is based on recording the spatial radiance distribution (SRD) at the surface of a sample. SRD provides information for measuring structural parameters of a fluorescent source (such as radius and depth). The aim of this article is to estimate the depth and radius of the source distribution from SRD, measured at the sample surface. For this reason, a theoretical expression for the SRD at the surface of a turbid sample arising from a spherical light source embedded in the sample, was derived using a steady-state solution of the diffusion equation with an appropriate boundary condition.MethodsThe SRD was approximated by solving the diffusion equation in an infinite homogeneous medium with solid spherical sources in cylindrical geometry. Theoretical predications were verified by experiments with fluorescent sources of radius 2-6 mm embedded at depths of 2-4 mm in a tissue-like phantom.ResultsThe experimental data were compared with the theoretical values which shows that the root mean square (RMS) error in depth measurement for nominal depth values d = 2, 2.5, 3, 3.5, 4 mm amounted to 17%, 5%, 2%, 1% and 5% respectively. Therefore, the average error in depth estimation was ≤ 4% for depths larger than the photon mean free path.ConclusionsAn algorithm is proposed that allows estimation of the location and radius of a spherical source in a homogeneous tissue-like phantom by accounting for anisotropic light scattering effect using FRI modality. Surface SRD measurement enabled accurate estimates of fluorescent depth and radius in FRI modality, and can be used as an element of a more general tomography reconstruction algorithm.

We use seismic modeling of shot records to simulate the effects of acoustic‐wave propagation, P/SV-wave conversion, intrabed multiple reflections, anelastic attenuation, and transverse isotropy in materials representative of a shallow Gulf of Mexico environment. The study focuses on the estimates of three kinematic properties: vertical traveltime, velocity used for depth conversion, and depth to the reflector. The primary tool used for the measurement is velocity spectral analysis along hyperbolic traveltime‐distance curves. Analyses of numerous modeling experiments in the Gulf of Mexico environment lead to the following conclusions: First, the kinematic properties are affected mainly by anelastic attenuation, vertical velocity heterogeneity, and velocity anisotropy. The effects of P/SV-wave mode conversion and intrabed multiples are of secondary significance. Second, anelastic attenuation coupled with heterogeneity contribute up to 3 percent in depth error, mainly in the form of apparent time delays. Transverse isotropy is likely to add up to an additional 2 percent, for a total of 5 percent depth error. Since the depth error magnitude agrees closely with published field observations, we conclude, therefore, that attenuation, heterogeneity, and transverse isotropy account for most of the errors in depth estimation in horizontally layered media.

Depth estimation of deep water moving source based on ray separation

While many stereo vision algorithms can quickly and robustly estimate sparse geometry from sets of photos, dense reconstruction, where depth estimate is required at per-pixel or sub-pixel level, remains a time-consuming and memory-intensive process. In this work, we propose a fast hierarchical upsampling method for dense image-based depth estimation. The main idea is to start from sparse depth estimates that can be quickly computed using any existing multiview stereopsis tool, then iteratively upsample the depth values to obtain a dense reconstruction consisting of millions of points. Using a GPU-based implementation, the upsampling algorithm can perform up to 15 images per second. The results can be directly used for 3D modeling applications; in addition, they can be used to digitally manipulate the depth-of-field effects in the input images in order to simulate refocusing.

Firstly, the elementary principle of passive estimation of target depth is studied, which is supported by underwater acoustic detection system of vertical linear array and is conducted in deep sea. Then, error sources of passive estimation of target depth are analysed theoretically from points of view of random error and deployment deviation of vertical linear array, and correlative formulae are deduced. After simulation experiments, it is concluded that how these error sources impact accuracy of passive estimation of target depth, which is the basis of further research for passive estimation of target depth and improvement of accuracy of passive estimation of target depth.

This paper presents a new target-depth estimation method that is based on ray back-propagation with a probabilistic approach. This localization algorithm tries to minimize the mean-squared error of elevation angles at the receiver and arrival times between a model and measures. This method was tested on Monte-Carlo simulations of classic active sonar scenarios and using experimental data from a real tank. In active sonar with a point-target model, combined acoustic paths may exist. These have a different path between the sonar-target and the target-array. This paper discusses also about this ray identification. Simulations with vertical array suggest that the target-depth estimation can be realized with a low uncertainty compared to the water column for long ranges in a Mediterranean sound speed profile. However, some environmental parameters as random sound-speed profile, array depth or array tilts could increase the bias and the variance of the target-depth estimator. Results on experimental data with surface noise reveal a good estimation of the target depth and validate our localization algorithm on a constant sound-speed profile with a vertical array.

Recent advances in deep convolutional neural network have lead to significant improvement over depth estimation from a single image. However, training such networks usually needs a large amount of high-quality labeled training data which is difficult to collect. To alleviate it, a semi-supervised method based on confidence learning is proposed to improve the results of depth estimation with additional unlabeled data. We utilize the confidence map generated by a confidence network to predict the trusty regions of depth estimations on unlabeled data. These depth estimations are taken as pseudo ground truth and used to train the depth network together with labeled data. Experiments on NYU Depth dataset V2 show our proposed semi-supervised method outperforms fully-supervised method.

Biases in the depth estimation of expendable bathythermograph (XBT) measurements cause considerable errors in oceanic estimates of climate variables. Efforts are currently underway to improve XBT probes by including pressure switches. Information from these pressure measurements can be used to minimize errors in the XBT depth estimation. This paper presents a simple method to correct the XBT depth biases using a number of discrete pressure measurements. A blend of controlled simulations of XBT measurements and collocated XBT/CTD data is used along with statistical methods to estimate error parameters, and to optimize the use of pressure switches in terms of number of switches, optimal depth detection, and errors in the pressure switch measurements to most efficiently correct XBT profiles. The results show that given the typical XBT depth biases, using just two pressure switches is a reliable strategy for reducing depth errors, as it uses the least number of switches for an improved accuracy and reduces the variance of the resulting correction. Using only one pressure switch efficiently corrects XBT depth errors when the surface depth offset is small, its optimal location is at middepth (around or below 300 m), and the pressure switch measurement errors are insignificant. If two pressure switches are used, then results indicate that the measurements should be taken in the lower thermocline and deeper in the profile, at approximately 80 and 600 m, respectively, with an RMSE of approximately 1.6 m for pressure errors of 1 m.

Ship target feature extraction plays an important role in ship recognition and tracking based on video surveillance. However, the current feature extraction of ship targets based on video surveillance does not have a complete summary, so this article reviews these methods. This article summarizes the main ship target feature extraction methods used in ship target recognition and tracking in different application environments, including traditional artificial ship feature extraction and automated learning ship feature extraction. It focuses on the core ideas, advantages and disadvantages of the representative methods in each type of problem, and discusses the current problems in the feature extraction of ship targets and the future development of feature extraction technology.

SLAM is one of the most critical technologies for mobile robots. Vision-based SLAM techniques have recently become popular. In this study, a high-density reconstruction technique based on monocular camera, MDF-SLMA, is proposed to solve the constraints of stereo and RGB-D cameras. The DSO-based tracking thread is used for RGB input images. To estimate the depth of key frames, an updated deep learning technique is used. Attitude optimization and density mapping are based on the results of depth estimation. Experimental results show that the proposed method can achieve monocular density mapping and reduce SLAM trajectory error. MDF-SLAM lays the foundation for mobile robots to detect their surroundings and perform complex tasks.

Converting tower-mounted videos from perspective to orthographic view is beneficial for their integration with maps and remote sensing images and can provide a clearer and more real-time data source for earth observation. This paper addresses the issue of low geometric accuracy in orthographic video generation by proposing a method that incorporates 3D GIS view matching. Firstly, a geometric alignment model between video frames and 3D GIS views is established through camera parameter mapping. Then, feature point detection and matching algorithms are employed to associate image coordinates with corresponding 3D spatial coordinates. Finally, an orthographic video map is generated based on the color point cloud. The results show that (1) for tower-based video, a 3D GIS constructed from publicly available DEMs and high-resolution remote sensing imagery can meet the spatialization needs of large-scale tower-mounted video data. (2) The feature point matching algorithm based on deep learning effectively achieves accurate matching between video frames and 3D GIS views. (3) Compared with the traditional method, such as the camera parameters method, the orthographic video map generated by this method has advantages in terms of geometric mapping accuracy and visualization effect. In the mountainous area, the RMSE of the control points is reduced from 137.70 m to 7.72 m. In the flat area, it is reduced from 13.52 m to 8.10 m. The proposed method can provide a near-real-time orthographic video map for smart cities, natural resource monitoring, emergency rescue, and other fields.