Three-dimensional Laser Imaging Research Articles

Automatic detection of potholes using VGG-16 pre-trained network and Convolutional Neural Network

A self-driving car is necessary to implement traffic intelligence because it can vastly enhance both the safety of driving and the comfort of the driver by adjusting to the circumstances of the road ahead. Road hazards such as potholes can be a big challenge for autonomous vehicles, increasing the risk of crashes and vehicle damage. Real-time identification of road potholes is required to solve this issue. To this end, various approaches have been tried, including notifying the appropriate authorities, utilizing vibration-based sensors, and engaging in three-dimensional laser imaging. Unfortunately, these approaches have several drawbacks, such as large initial expenditures and the possibility of being discovered. Transfer learning is considered a potential answer to the pressing necessity of automating the process of pothole identification. A Convolutional Neural Network (CNN) is constructed to categorize potholes effectively using the VGG-16 pre-trained model as a transfer learning model throughout the training process. A Super-Resolution Generative Adversarial Network (SRGAN) is suggested to enhance the image's overall quality. Experiments conducted with the suggested approach of classifying road potholes revealed a high accuracy rate of 97.3%, and its effectiveness was tested using various criteria. The developed transfer learning technique obtained the best accuracy rate compared to many other deep learning algorithms.

Macrotexture deterioration for micromilled tunnel concrete pavement using 3D laser data

To ensure driving safety and cost effectively preserve pavements, unlike the traditional grooving, cutting joints, highway tunnels used micromilling resurfacing method to improve the skid resistance of concrete pavement. The sensor measured texture depth (SMTD) has been established for quantitatively evaluating micromilled surface quality. This paper explores the possibility of using three-dimensional line laser imaging technology to obtain the ‘full lane’ elevation data of the pavement, then through the calculation of the index SMTD, construct a road surface spatial and temporal deterioration model, and provide quality control covering the whole lane, and identify the problem areas. The results show that the road wear condition gradually worsens with the increase in opening time, after 24 months, the proportion of the good level (>0.55 mm) of the three cells is 70.5%, 75.1%, and 70.2%. Analysis of the test data for three cells revealed that the tunnel pavement shows canalization in wear. The most severely abraded section is offset by 200–300 mm toward the roadway median compared to the theoretical vehicle wheel track zone. Subsequent studies can expand the experimental conditions and consider more influencing factors to study further the texture deterioration of the micromilled pavement.

Three-Dimensional Image Reconstruction With Enhanced Spatial Accuracy for Pulsed-Laser Imaging Based on Compressive Sensing

Compressive sensing (CS) has been intensively studied as a superresolution technique in image acquisition and reconstruction. It has also been introduced into three-dimensional (3-D) laser imaging to enhance the spatial resolution of images obtained from existing few-pixel array sensors. These approaches, however, either suffer from low range resolution or lack long-range detecting abilities. To solve the CS reconstruction problem under the 3-D pulsed-laser imaging framework, we build a sampling model to combine the CS sampling process with the time-of-flight-based range measurement procedure in pulsed-laser imaging. To explore range information from the overlapping echo signals, we develop a gradual alternating minimization algorithm with a designed range equivalent matrix, in which the feasible region of the solution gradually shrinks to make the reconstruction procedure tractable. To further enhance the quality of the reconstructed intensity image, we introduce rank minimization and a fallback mechanism, in which the sparsity and stability of the reconstructed intensity image is refined. In this manner, the accuracy of reconstructed 3-D images can be improved in both the pixel direction and range direction. The simulation results based on real datasets demonstrate that our proposed scheme can yield superior accurate reconstruction and leads to significant improvements compared with recent approaches.

Digital Evaluation of Nitrite-Reduced “Kulen” Fermented Sausage Quality

This study aimed to evaluate nitrite reduction impact on geometry, colour, chemical, microbiological, and sensory traits of dry sausage (kulen) traditionally prepared with red hot paprika powder. Three batches of kulen with different nitrite levels were produced and assessed: N110 (control with 110 mg/kg of sodium nitrite), N55 (55 mg/kg of sodium nitrite), and NF (without sodium nitrite). Samples for the analyses were taken on production day, after 8, 16, 24, 32, and 40 days of ripening and after 50 and 100 days of storage. Four novel digital methods for quality assessment were deployed such as computer vision system (CVS), three-dimensional (3D) laser imaging, oral processing, and temporal dominance of sensations (TDS). Reduction and removal of nitrites from the formulation of kulen did not result in significant ( P < 0.01 ) differences in lightness ( L ∗ ), redness ( a ∗ ), and yellowness ( b ∗ ) of the sausage surface, meat, and fat parts that were measured independently by means CVS. Sausages produced by 50% nitrite reduction (N55) showed no significant ( P < 0.01 ) differences in terms of geometrical, chemical, colour, microbiological, and oral processing parameters compared with the control (N110) batch. On the other hand, the complete removal of nitrites from kulen formulation negatively affected biogenic amine levels and oral processing properties of the product. Nitrite reduction showed no significant effect on TDS curves among the batches. The results of this study indicate that nitrite content in traditional kulen can be reduced by 50% (55 mg/kg of sodium nitrite) without adversely affecting the various quality properties of the product.

Reformulation of Traditional Fermented Tea Sausage Utilizing Novel (Digital) Methods of Analysis

The main objective of this paper was to investigate the effect of fat reduction on different quality traits of tea sausage. This study also aimed to deploy the following digital methods of analysis: three-dimensional (3D) laser imaging, computer vision system and oral processing. Three batches of tea sausage with different amounts of pork back fat were manufactured: control (25%), medium fat (17.5%) and low fat (10%). Samples for the analyses were taken on the production day and after 7, 14, 21, 28 and 35 days of ripening. The fat level significantly influenced shrinkage, texture, pH, aw, moisture and ash contents, peroxide value, acid number, number of chewing strokes, consumption time, eating rate and fat intake rate. Oxidative stability, colour and microbiological parameters were not affected by fat reduction. The results of the sensory analysis showed that the fat level can be reduced to 17.5% without negatively affecting the quality and sensory properties of the product. The ripening time of the fat-reduced tea sausage should be reduced to 28 days. A strong correlation between shrinkage and weight loss suggests the possibility of using 3D laser imaging in predicting weight loss and moisture content of dry sausages.

Three-Dimensional Laser Imaging with a Variable Scanning Spot and Scanning Trajectory

Traditional lidar scans the target with a fixed-size scanning spot and scanning trajectory. Therefore, it can only obtain the depth image with the same pixels as the number of scanning points. In order to obtain a high-resolution depth image with a few scanning points, we propose a scanning and depth image reconstruction method with a variable scanning spot and scanning trajectory. Based on the range information and the proportion of the area of each target (PAET) contained in the multi echoes, the region with multi echoes (RME) is selected and a new scanning trajectory and smaller scanning spot are used to obtain a finer depth image. According to the range and PAET obtained by scanning, the RME is segmented and filled to realize the super-resolution reconstruction of the depth image. By using this method, the experiments of two overlapped plates in space are carried out. By scanning the target with only forty-three points, the super-resolution depth image of the target with 160 × 160 pixels is obtained. Compared with the real depth image of the target, the accuracy of area representation (AOAR) and structural similarity (SSIM) of the reconstructed depth image is 99.89% and 98.94%, respectively. The method proposed in this paper can effectively reduce the number of scanning points and improve the scanning efficiency of the three-dimensional laser imaging system.

Effect of aggregate properties on asphalt pavement friction based on random forest analysis

Pavement friction largely depends on surface texture characteristics, which are directly related to the corresponding aggregate properties. This study selected forty-five field pavement sites, which were constructed using seven typical types of aggregates available from eleven different quarries in Oklahoma, as the testing beds. Aggregate properties were characterized in the laboratory using the Aggregate Imaging System (AIMS), and pavement texture and friction data were collected in parallel in the field using a three-dimensional laser imaging system and Grip Tester. The traditional multivariate friction models were developed while several regression coefficients did not follow engineering judgment. Subsequently, the random forest analysis (RFA) was performed to evaluate the pavement surface friction subject to various aggregate properties. Loss of texture (%) is identified as the most significant aggregate characteristic for friction evaluation, accounting for 26.54% contribution to pavement surface friction. Traffic volume and temperature also exhibit significant impacts, with 31%, 14.7% contributions to pavement surface friction, respectively. This study is beneficial to better understand the effect of aggregate properties on pavement friction and to improve skid resistance performance in the field.

Aggregate Characteristics-Based Preventive Maintenance Treatments for Optimized Skid Resistance of Pavements

Skid resistance of pavements plays a significant role in roadway safety. Preventive maintenance (PM) treatments have been used to restore pavement condition and retard future deterioration. As aggregate is the main component of such treatments, this study assessed the skid resistance of pavements in Oklahoma following PM treatment application to evaluate whether skid resistance is maintained or improved and economics are optimized, and presents a methodology to assist pavement engineers manage skid resistance. The 13 commonly used sources of aggregates in Oklahoma were acquired for aggregate morphological characterization in the laboratory in relation to shape, angularity, and surface texture-related index properties, using the Aggregate Imaging System before and after Micro-Deval abrasion. Accordingly, a systematic experimental design was developed to include 45 field testing sites constructed with typical Oklahoma PM treatments and the common aggregate sources. Multiple rounds of field data collection were performed during a 3-year period using the most recent three-dimensional (3D) laser imaging technology for pavement surface characteristics and a Grip Tester for skid resistance. The impacts of aggregate properties and pavement surface characteristics on the skid resistance were investigated and used to develop statistical deterioration models. Subsequently, life cycle cost analysis and performance-based sensitivity analysis were conducted using the developed deterioration model. This study presents a detailed analysis of aggregate characteristics and its relationship to skid resistance of pavements. An understanding of these relationships and application of the presented methodology can result in improvements in pavement safety with enhanced stewardship of financial resources.

Frequency-multiplexing photon-counting multi-beam LiDAR

We report a frequency-multiplexing method for multi-beam photon-counting light detection and ranging (LiDAR), where only one single-pixel single-photon detector is employed to simultaneously detect the multi-beam echoes. In this frequency-multiplexing multi-beam LiDAR, each beam is from an independent laser source with different repetition rates and independent phases. As a result, the photon counts from different beams could be discriminated from each other due to the strong correlation between the laser pulses and their respective echo photons. A 16-beam LiDAR system was demonstrated in three-dimensional laser imaging with 16 pulsed laser diodes at 850 nm and one single-photon detector based on a Si-avalanche photodiode. This frequency-multiplexing method can greatly reduce the number of single-photon detectors in multi-beam LiDAR systems, which may be useful for low-cost and eye-safe LiDAR applications.

Layered compressive sensing reconstruction for non-scanning three-dimensional laser imaging

ABSTRACTAs an active remote-sensing technique, three-dimensional (3D) laser radar enables accurate target detection and identification. However, the resolution of obtained image is limited by the integrated level of the sensor array. In this article, we propose a layered compressive sensing reconstruction method for high-resolution 3D laser imaging. The range profile is first reconstructed by applying a designed range observation matrix to the original sampling data. According to the distribution of range values in the reconstructed range profile, the original sampling data are divided into several layers and the corresponding intensity images are reconstructed by different algorithms. The global intensity image is generated by the reconstructed intensity images with different range compensation coefficients. In our simulations based on real data from the Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Model Version 2 (ASTER GDEM V2) and Landsat-8 Operational Land Imager (OLI), the proposed method realised 3D laser imaging for complex landforms with a pixel resolution of , and reached a satisfactory performance for reconstructions of the range profile (peak signal-to-noise ratio (PSNR) = 33.08 dB, feature similarity (FSIM) = 0.9450) and the intensity image (PSNR = 20.63 dB, FSIM = 0.8893) with a subrate of 0.25.

Compressive Sensing-Based Three-Dimensional Laser Imaging With Dual Illumination

Low-cost 3D imaging, particularly by using laser detection and ranging (LIDAR), is important for applications such as object recognition, surface mapping, and machine vision. Conventional time-of-flight LIDAR uses a scanned laser to obtain the intensity and range of targets, which requires a narrow bandwidth of illumination and high-speed synchronizers. This paper presents a nonscanning prototype of a pulse-width-free 3D LIDAR which combines single-pixel imaging and diffractive optical elements, for the first time to our knowledge. Compressive sensing techniques are used to measure echo pulses from the target and reconstruct the intensity map of the target scene. Diffractive optical elements are also applied to generate structured illumination and the depth map of the target scene can be obtained from laser spot extraction. The simulation results are presented to verify the effectiveness of the proposed prototype as well as illustrate its superiority where traditional 3D imaging methods are unavailable or limited. This novel prototype has advantages of low cost and flexible structure at wavelengths beyond the visible spectrum and will be highly interesting for practical applications.

Subpixel three-dimensional laser imaging with a downscaled avalanche photodiode array using code division multiple access

Scannerless time-of-flight three-dimensional imaging, the successor to raster scanning three-dimensional imaging, relies on large-scale detector arrays to obtain high pixel resolution; however, manufacturing limitations lead to a bottleneck in imaging resolution. Here, we report a methodology for laser imaging using code division multiple access, which involves three key steps. Optical encoding is carried out for a pulsed laser to generate space-time encoded beams for projection. Optical multiplexing subdivides the backscattered light signals and multiplexes them to a downscaled avalanche-photodiode array. Subpixel decoding decodes the digitized encoded full waveforms and decomposes the features of all subpixels. Accordingly, we design a prototype based on a 64-order encoder and a 4-element avalanche-photodiode array and conduct an outdoor experiment. We demonstrate that the system is capable of obtaining 256 pixels per frame in push-broom mode and reconstruct a three-dimensional image with centimetre-level lateral resolution and range precision at a distance of ∼112 m.

A new three-dimensional nonscanning laser imaging system based on the illumination pattern of a point-light-source array.

One of the most important goals of research on three-dimensional nonscanning laser imaging systems is the improvement of the illumination system. In this paper, a new three-dimensional nonscanning laser imaging system based on the illumination pattern of a point-light-source array is proposed. This array is obtained using a fiber array connected to a laser array with each unit laser having independent control circuits. This system uses a point-to-point imaging process, which is realized using the exact corresponding optical relationship between the point-light-source array and a linear-mode avalanche photodiode array detector. The complete working process of this system is explained in detail, and the mathematical model of this system containing four equations is established. A simulated contrast experiment and two real contrast experiments which use the simplified setup without a laser array are performed. The final results demonstrate that unlike a conventional three-dimensional nonscanning laser imaging system, the proposed system meets all the requirements of an eligible illumination system. Finally, the imaging performance of this system is analyzed under defocusing situations, and analytical results show that the system has good defocusing robustness and can be easily adjusted in real applications.

Multihypothesis-Based Compressive Sensing Algorithm for Nonscanning Three-Dimensional Laser Imaging

The resolution of nonscanning three-dimensional (3-D) imaging systems is limited by the number and accuracy of the array sensors. Moreover, for space-continuous targets in practical situations, the echo pulses in time-of-flight systems overlap and traditional peak detection is no longer suitable for super-resolution applications. Hence, compressive sensing (CS) has been introduced to achieve super-resolution. However, most of the conventional CS algorithms cannot be used directly for 3-D image reconstruction. In this paper, we propose a novel super-resolution algorithm for nonscanning 3-D laser imaging based on CS reconstruction. To acquire the range information of space-continuous targets, an all-one projection is implemented in advance to estimate the spatial distribution of the targets; a range observation matrix composed of time-interval basis vectors is then constructed to obtain the peak values of each frame from overlapping echo pulses. Because of the spatial continuity of the targets, Tikhonov regularization is utilized to solve the ill-posed inverse problem. Furthermore, to enhance the reconstruction quality of the adjacent frames, multihypothesis prediction is used with displacement and diffusion models to estimate the motion of the contour line. Simulation results based on real data from the ASTER global digital elevation model demonstrate the effectiveness and high accuracy of the proposed algorithm for complex landforms.

Design and modeling of three-dimensional laser imaging system based on streak tube.

In the streak tube laser imaging system, there are two conflicts: the first is between high spatial resolution and wide field of view (FOV). The second is between high temporal resolution and deep depth of field (DOF). In this paper, a new non-scanning streak tube laser imaging system is presented. A microlens array with three different apertures is used to non-uniformly collect data from the image plane so the conflict between high spatial resolution and wide FOV is relieved and the detectable range of system is also increased. A remapping fiber optics with special design is used to realign the image plane on the two photocathodes of streak tubes to realize the operation mode of the two streak tubes so the conflict between high temporal resolution and deep DOF is relieved. The mathematical model of the entire imaging system is established based on the range equation. The structure parameters of the receiving optical system are optimized in order to achieve the optimal utilization rate of light energy. In the third, three simulated contrast experiments are organized, and the experiment results demonstrate that the imaging system proposed in this paper possesses properties of higher spatial resolution, wider FOV, higher temporal resolution, deeper DOF, and larger detectable range.

Underwater three-dimensional imaging laser sensor with 120-deg wide-scanning angle using the combination of a dome lens and coaxial optics

We developed an underwater three-dimensional (3-D) imaging sensor using a 532-nm laser. The sensor system combines a dome lens with coaxial optics to realize a wide-scanning angle of 120 deg (horizontal)×30 deg (vertical) while having a compact size of 25-cm diameter and 60-cm length. A detector sensitivity time control circuit and a time-to-digital converter are used to detect a small signal and suppress the unwanted backscattered signals due to marine snow. 3-D imaging of the seafloor with 20-m width and 60-m length was demonstrated in the sea around Ishigaki Island, Japan.

A compressive sensing algorithm using truncated SVD for three-dimensional laser imaging of space-continuous targets

For traditional array 3-D laser radars, the resolution of the intensity image and range profile is limited by the number and accuracy of sensors. Moreover, for a space-continuous target, peak detection in the pulsed time of flight is no longer suitable for super-resolution reconstruction algorithms. Hence, a compressive sensing algorithm for 3-D laser imaging is proposed. A range observation matrix composed of time interval basis vectors is constructed to acquire the range information regarding a target. However, the range observation matrix is generally ill-posed owing to the spatial continuity of the target. To address this shortage, truncated singular value decomposition is utilized to extract the peak values of echo pulses for image reconstruction. Simulation results demonstrate the effectiveness and performance of the proposed algorithm.

Modeling and simulations of three-dimensional laser imaging based on space-variant structure

A three-dimensional (3D) laser imaging system based on time of flight is proposed, based on the human retina structure. The system obtains 3D images with space-variant resolution, and we further establish mathematical models of the system and carried out simulative comparison between space-variant structure (SVS) and space-invariant structure (SIS). The system based on SVS produces significant improvements over traditional system based on SIS in the following aspects: (1) The system based on SVS uses less pixels than that based on SIS under the same field of view (FOV) and resolution. Therefore, this property is more suitable for uses in situations that require high speed and large volume data processing. (2) The system based on SVS has higher efficiency of utilization of photodiode array than that based on SIS. (3) 3D image based on SVS has the properties of rotation and scaling invariance. (4) The system based on SVS has higher echo power in outside ring of large photodiode array, which is more effective in detecting targets with low reflectance.

Three-dimensional imaging reconstruction algorithm of gated-viewing laser imaging with compressive sensing.

Three-dimensional (3D) laser imaging combining compressive sensing (CS) has an advantage in lower power consumption and less imaging sensors; however, it brings enormous stress to subsequent calculation devices. In this paper we proposed a fast 3D imaging reconstruction algorithm to deal with time-slice images sampled by single-pixel detectors. The algorithm implements 3D imaging reconstruction before CS recovery, thus it saves plenty of runtime of CS recovery. Several experiments are conducted to verify the performance of the algorithm. Simulation results demonstrated that the proposed algorithm has better performance in terms of efficiency compared to an existing algorithm.

Development of an algorithm to measure defect geometry using a 3D laser scanner

Current fatigue life prediction models for coiled tubing (CT) require accurate measurements of the defect geometry. Three-dimensional (3D) laser imaging has shown promise toward becoming a nondestructive, non-contacting method of surface defect characterization. Laser imaging provides a detailed photographic image of a flaw, in addition to a detailed 3D surface map from which its critical dimensions can be measured. This paper describes algorithms to determine defect characteristics, specifically depth, width, length and projected cross-sectional area. Curve-fitting methods were compared and implicit algebraic fits have higher probability of convergence compared to explicit geometric fits. Among the algebraic fits, the Taubin circle fit has the least error. The algorithm was able to extract the dimensions of the flaw geometry from the scanned data of CT to within a tolerance of about 0.127 mm, close to the tolerance specified for the laser scanner itself, compared to measurements made using traveling microscopes. The algorithm computes the projected surface area of the flaw, which could previously only be estimated from the dimension measurements and the assumptions made about cutter shape. Although shadows compromised the accuracy of the shape characterization, especially for deep and narrow flaws, the results indicate that the algorithm with laser scanner can be used for non-destructive evaluation of CT in the oil field industry. Further work is needed to improve accuracy, to eliminate shadow effects and to reduce radial deviation.