Frame Size Research Articles

Digital Image Decoder for Efficient Hardware Implementation

Increasing the resolution of digital images and the frame rate of video sequences leads to an increase in the amount of required logical and memory resources necessary for digital image and video decompression. Therefore, the development of new hardware architectures for digital image decoder with a reduced amount of utilized logical and memory resources become a necessity. In this paper, a digital image decoder for efficient hardware implementation, has been presented. Each block of the proposed digital image decoder has been described. Entropy decoder, decoding probability estimator, dequantizer and inverse subband transformer (parts of the digital image decoder) have been developed in such way which allows efficient hardware implementation with reduced amount of utilized logic and memory resources. It has been shown that proposed hardware realization of inverse subband transformer requires 20% lower memory capacity and uses less logic resources compared with the best state-of-the-art realizations. The proposed digital image decoder has been implemented in a low-cost FPGA device and it has been shown that it requires at least 32% less memory resources in comparison to the other state-of-the-art decoders which can process high-definition frame size. The proposed solution also requires effectively lower memory size than state-of-the-art architectures which process frame size or tile size smaller than high-definition size. The presented digital image decoder has maximum operating frequency comparable with the highest maximum operating frequencies among the state-of-the-art solutions.

Orthogonal Time Frequency Space Detection via Low-Complexity Expectation Propagation

Orthogonal time frequency space (OTFS) can provide better error performance than orthogonal frequency division modulation in the high-speed scenario. However, the two-dimensional convolution between the information-bearing symbols and the channel response in the delay-Doppler domain induces high-complexity detection, which hinders the implementation of the OTFS. In this paper, expectation propagation (EP) is introduced in the OTFS system to improve the reliability compared to the message passing approaches. The low-complexity EP detection with a log-linear order is proposed by observing the sparsity and the block quasi-banded matrix structure of the time-domain matrix. The exploitation of the lower-upper factorization for the banded matrix, the forward or the back substitution algorithm for the lower or the upper triangular matrices reduces the complexity of the large-scale matrix inversion involved in the EP detection from a cubic order to a linear order. The error performance of the proposed scheme is further analyzed by utilizing the state evolution and discussed under the different frame sizes. Moreover, the proposed low-complexity detector is then explored together with the likelihood decoder in an iterative manner to further improve the reliability of the proposed receiver. Finally, the better error performance and the lower computational complexity of the proposed receiver are validated by the numerical results.

Grouping based radio frequency identification anti-collision protocols for dense internet of things application

<span lang="EN-US">Radio frequency identification (RFID) is an important internet of things (IoT) enabling technology. In RFIDs collision occur among tags because tags share communication channel. This is called tag collision problem. The problem becomes catastrophic when dense population of tags are deployed like in IoT. Hence, the need to enhance existing dynamic frame slotted ALOHA (DFSA) based electronic product code (EPC) C1G2 media access control (MAC) protocol. Firstly, this paper validates through simulation the DFSA theory that efficiency of the RFID system is maximum when the number tags is approximately equal to the frame size. Furthermore, literature review shows tag grouping is becoming popular to improving the efficiency of the RFID system. This paper analyzes selected grouping-based algorithms. Their underlining principles are discussed including their tag estimation methods. The algorithms were implemented in MATLAB while extensive Monte Carlo simulation was performed to evaluate their strengths and weaknesses. Results show that with higher tag density, fuzzy C-means based algorithm (FCMBG) outperformed traditional DFSA by over 40% in terms of throughput rate. The results also demonstrate FCMBG bettered other grouping-based algorithms (GB-DFSA and GBSA) whose tag estimation method are based on collision slots in terms slot efficiency by over 10% and also in terms of identification time.</span>

Zoning of the reaction of the tissues of the implantation area after prosthetic repair in abdominal hernias in experiment

Purpose of the study. Development of criteria for evaluating the reaction of tissues in the implantation area after prosthetic repair of the hernia ring in an experiment using new digital image processing technologies.Materials and methods. The experiment was carried out on 13 Wistar rats of different sexes. Under anesthesia, a 4 × 2 cm mesh prosthesis was implanted in the layer similar to the onlay epiaponeurotic. The preparation and staining of each preparation was carried out in two versions: with hematoxylin-eosin and according to Mallory on different glasses. The seasons for performing experiments are winter, summer. Microphotography was performed on a medical transmitted light microvisor Microvisor μVizo‑103 (AD LOMO, Russia) with a 3.2 Mpx matrix (full frame size 1024 × 882 px, infoband width 88 px) with Planachromat and Stigachromat lenses with a digital scale × 1 (i.e. no digital zoom). Microphoto processing was performed using the ImageJ 1.53k program (version for public use). The criterion for including a microphoto in the calculation of indicators was the presence of an ingrown implant fiber in the preparation or a cavity surrounding the implant.Results. The zone adjacent to the implant, with a width of about 50 µm ("the zone of changes in adjacent tissues") significantly differs from the "zone of changes in distant tissues" by a more pronounced inflammatory reaction due to a larger absolute number of lymphocytes (р = 0.019 and fibroblasts (р = 0.017), giant cells of foreign bodies (р < 0.001), larger vascular area (р < 0.001) and extravasations (р = 0.002). The study of the "zone of changes in adjacent tissues" is of key importance in assessing the local response of the body to implantation. The key zones of the implantation area are identified: "the cavity surrounding the implant", "the zone of changes in adjacent tissues", "the zone of changes in distant tissues". The calculation of indicators within these zones allows you to get more detailed results when comparing study groups in the experiment.Conclusion. An effective method for assessing the reaction of tissues in the implantation area after prosthetic hernia repair in an experiment using new digital imaging technologies that can be used using a conventional optical microscope and standard staining methods is proposed.

Optimising large-area crystal orientation mapping of nanoscale β phase in α + β titanium alloys using EBSD

α + β titanium alloys, such as the commercially important alloy Ti-6Al-4 V (Ti64), develop complex textures during thermomechanical processing due to the allotropic phase transformation between the β and α phases. These phases are crystallographically related through the Burgers orientation relationship (BOR) and are often characterised by electron backscatter diffraction (EBSD) in the scanning electron microscope (SEM). However, the BOR can be destroyed for the primary α in wrought Ti64, and techniques that utilise the BOR to reconstruct the β phase from the room temperature α-phase data cannot be used. Instead, the β texture must be measured from the residual, nanoscale β ligaments in the room temperature microstructure, which are challenging to index because of orientation and phase overlap in the EBSD Kikuchi patterns. In this work, the SEM-EBSD acquisition and processing parameters were systematically varied to determine how best to index the residual β in Ti64, and an experimental methodology was thus developed to measure the β-phase texture efficiently and reliably after thermomechanical processing. The best compromise for maximising indexing of the residual β was achieved with a low current (~1 nA in this case), and additional indexing was achieved by increasing the sample stage tilt (to 75° in this case) and by limited the scanning frame size. It was shown that the β-phase texture could be reliably measured with the optimised beam and tilt settings using a relatively coarse step size (3 μm), but this approach does not yield any morphological or spatially relevant microstructure data. Thus, it was proposed that numerous multi-scale scans be performed with different settings to characterise the residual β phase in T64, each optimised to acquire either bulk texture analysis or microstructure spatial and morphological detail. • Residual β in Ti64 can be detected with EBSD using carefully selected parameters. • A ~ 1 nA current was the best compromise for maximising β indexing and EBSP intensity. • Additional β indexing was also achieved by increasing the sample tilt to 75°. • Dictionary Indexing can use biased pattern solutions to boost β-phase indexing.

Machine Learning based Frame Size Optimization in WLAN Downlink MU-MIMO Channel with the Least Cost of Delay

The key enhancement in the medium access control (MAC) layer is frame aggregation introduced by the IEEE 802.11n/ac standard to accommodate the growing traffic demand in the WLAN by allowing multiple packets aggregated per transmission. Frame aggregation efficiently reduces control overhead in the MAC layer, such as the MAC header and thus it helps to enhance transmission efficiency and throughput performance of WLAN. However, heterogeneous traffic demand among streams in the WLAN downlink MU-MIMO channel creates a challenge to efficiently utilize the benefits of frame aggregation. Transmission efficiency is also compromised during frame size setting determination because when a frame size is larger, the impact of the overhead frame can be lower, but they are also more vulnerable to transmission errors. Thus, this trade-off between maximizing frame size and minimizing overhead frames should be addressed by employing an adaptive frame aggregation technique to derive the optimal frame size that would maximize the throughput in WLAN downlink MU-MIMO channel. Moreover, when frame aggregation approach is employed, more frames must wait before transmission in a buffer which causes a delay in the performance of WLAN. Thus, analysing the trade-off between maximizing throughput and minimizing delay is a critical issue that should also be addressed to enhance the performance of WLAN. However, the majority of the existing adaptive aggregation algorithms in the WLAN downlink MU-MIMO channel are focused to maximize the throughput or minimize the delay. The main contribution of this paper is to propose a machine learning-based frame size optimization algorithm by extending our earlier approach in considering the cost of delay to maximize the system throughput of WLAN. The effectiveness of the proposed scheme is evaluated over the FIFO Baseline Approach and earlier conventional approaches under the effects of various traffic patterns, channel conditions, and the number of STAs.

VToonify

Generating high-quality artistic portrait videos is an important and desirable task in computer graphics and vision. Although a series of successful portrait image toonification models built upon the powerful StyleGAN have been proposed, these image-oriented methods have obvious limitations when applied to videos, such as the fixed frame size, the requirement of face alignment, missing non-facial details and temporal inconsistency. In this work, we investigate the challenging controllable high-resolution portrait video style transfer by introducing a novel VToonify framework. Specifically, VToonify leverages the mid- and high-resolution layers of StyleGAN to render high-quality artistic portraits based on the multi-scale content features extracted by an encoder to better preserve the frame details. The resulting fully convolutional architecture accepts non-aligned faces in videos of variable size as input, contributing to complete face regions with natural motions in the output. Our framework is compatible with existing StyleGAN-based image toonification models to extend them to video toonification, and inherits appealing features of these models for flexible style control on color and intensity. This work presents two instantiations of VToonify built upon Toonify and DualStyleGAN for collection-based and exemplar-based portrait video style transfer, respectively. Extensive experimental results demonstrate the effectiveness of our proposed VToonify framework over existing methods in generating high-quality and temporally-coherent artistic portrait videos with flexible style controls. Code and pretrained models are available at our project page: www.mmlab-ntu.com/project/vtoonify/.

DESIGN AND ANALYSIS OF THE FRAME STRENGTH OF WASTE TRANSPORTING MACHINE USING SOLIDWORKS

Garbage is a problem of concern worldwide. Public awareness in disposing of waste that is not in its place causes various kinds of problems, one of the problems that occurs is in rivers or artificial canals. This is what makes many people create tools to make it easier to control improper waste disposal. One of the tools that was created was a garbage collector, in making the design of a garbage collection machine using solidworks software, Galvanized Steel material was used with a rod size of 30 x 30 x 2.6 (mm) as the frame. With a frame size of 1320 mm in length, 790 mm in width and 1133 mm in height. The results of the simulation using solidworks software on the frame of the garbage collection machine are to get the von misses stress value with a maximum stress value (MPa) and a minimum stress value of (MPa), for the displacement value with a maximum stress value of 0.61 mm and for a minimum stress value of 0.00 mm, the safety factor gets a value of 3.55 ul. In the comparison of solidworks and theoretical software values, the value for von misses stress is 0.04%, for displacement is 0.24%, and for safety factor is 0.20%.

Non-destructive observation of internal structures of epoxy monolith and co-continuous network polymer using X-ray CT imaging for elucidation of their unique mechanical features and fracture mechanism

New kind of polymer composites, i.e., co-continuous network polymers (CNP), were fabricated by the filling of the continuous pores of an epoxy monolith (EM) with a cross-linked acrylate polymer. The internal structures of EM and CNP were non-destructively observed by X-ray CT imaging. The mechanical properties of EM were highly dependent on the pore and epoxy frame sizes. Increasing the molar ratio of the amino groups to glycidyl groups during the monolith fabrication process created a coarser co-continuous structure. It greatly enhanced the stretching property of EM. Filling the pores of EM with a rubbery cross-linked acrylate polymer as the second component increased the maximum strength of CNP, independent of the pore size. Furthermore, the gap formation at the boundary between the hard phase and the soft phase of CNP was observed by X-ray CT imaging for the broken test piece used for a tensile test. We discussed a mechanism for the deformation and fracture of EM and CNP.

Object Video Tracking using a Pan-Tilt-Zoom System

This paper implements object video tracking system that represents object location in subsequent video frames. A closed-circuit television (CCTV) camera mounted on a rotating system is used for capturing video while the object of interest always kept at the centre of the screen. Three tracking algorithms were selected and implemented: template matching, contour matching and optical flow. Measures of their accuracy and speed were taken for comparison. The software was implemented in a personal computer with C# programming language, with the aid of EmguCV which is a wrapper for OpenCV, a famous image processing library. The system implemented for this study is able to successfully track a rigid body, discernible from the background objects with size up to 400×300 pixel for the Phase Alternate Line (PAL) system of 720×576 pixel frame size. Tracking was stable even with the existence of rotation and scaling. Some faults were observed when occlusion was present or when the target was moving with a speed faster than that of the rotation system of 30 degrees/s horizontal and 15 degrees/s vertical.

VOTA: A Heterogeneous Multicore Visual Object Tracking Accelerator Using Correlation Filters

A correlation filter (CF) is a lighter and faster solution for visual object tracking (VOT) compared with a deep neural network (DNN). However, the CF-based trackers introduce diverse computational kernels and require FP16 computations. To address these challenges, we present a VOT accelerator (VOTA), a domain-specific accelerator for CF-based VOT that meets real-time performance at high efficiency, while providing flexibility and programmability. The VOTA encompasses a Winograd convolution core (WINO), a fast Fourier transformation (FFT) core (FFT), and a vector core (VEC) for diverse kernels, integrated in a high-bandwidth star-ring topology. The VOTA’s frame-based instruction set and execution enable a 537 GFLOPS performance, adapt to variances of CF trackers, and reduce the code size. An instruction-chaining mechanism permits inter-core pipelining and improves the hardware utilization up to 84.2%. A 10.2-mm 2 28-nm FP16 system on chip (SoC) prototype incorporating the VOTA, an RISC-V host CPU, and other supportive peripherals is taped out and measured to achieve 2.45 TFLOPS/W at 0.72 V. Running the oriented particle CF (OPCF), a CF tracker enhanced by adaptive boosting and particle filtering, the SoC chip achieves 1157 frames/s ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$640\times480$ </tex-math></inline-formula> frame size) at 0.9 V and 175 MHz, consuming 296 mW.

Structural Behavior of steel frame Infilled with Ferrocement Precast Panel S ubjected to Lateral Load

Background: When a severe earthquake ground motion hits the structure even if it doesn’t collapse it may undergoes inelastic deformations. Steel structure infilled with precast panels becomes less efficient when subjected to large lateral loads like strong wind and earthquake. Hence assessment of actual performance of steel frames provided with infills in the event of earthquake is essential. Methods: In the current study, a prototype model of steel frame size 4m x4m bay & 4m height with G+2 story is scaled down and subjected to lateral seismic load by conducting shake table test to assess modal frequencies, the same structure is modelled in Hypermesh and analyzed in Nastran, ETABS software. The results obtained are validated for a 16X16m bay of 4m height commercial complex of G+9 Multi-storeyed structure infilled with Ferrocement Precast panels analysed in ETABS software for base shear, response spectrum and pushover analysis to assess the performance of the structure. Findings: The frequency from modal analysis was found to be 45.21 Hz, for G+2 storied model in ETABS, values obtained for frequency, base shear, performance point of the structure are in appreciable. Novelty: As per the traditional earthquake design methods, structures are designed for less forces than maximum considered for earthquake. The overall performance of the structure is found to be safe even infilled with Ferrocement precast panels. Advantages: Ferrocement panels as infills has benefit as they are slender from structural point of view. Compared with steel bare frame dynamic behavior of bare steel frame which is infilled with Ferrocement precast panel subjected to lateral loads is high that helps in assessing the damage in the event of an earthquake in critical zone V. Keywords: Infill; Ferrocement precast panels; Pushover Analysis; Performance point; Response spectrum

Test System Development and Experimental Study on the Fatigue of a Full-Scale Steel Catenary Riser

This paper presents a full-scale deep-water steel catenary riser fatigue test system. The proposed system can carry out fatigue tests on steel catenary risers, hoses, and subsea pipelines up to 21 m in length, ranging from 8 to 24 inches in diameter. The test system was realized by mechanical loading with loading control systems, and could carry out axial tension and compression, bending moment, torsion, and internal pressure to simulate all load types on deep-water steel catenary risers or subsea pipelines. The counterforce was sustained by a counterforce frame. Through mechanical simulation analysis, the authors determined the size of the counterforce frame and designed the connection form of the counterforce frame and loading system. According to the required loading capacity, the appropriate cylinder thickness and diameter were obtained through calculation. After the design and construction of the test system, the authors designed a fatigue test to confirm the loading capacity and accuracy of the test system. The authors performed full-scale testing to assess the fatigue performance of pipe-to-pipe mainline 5G girth welds fabricated to BS 7608. This test was designed according to the stress level of pipelines in the Lingshui 17-2 gas field, and the test results were compared with the calculation results of the S–N curve.

Expression of Beef- Versus Dairy-Type in Crossbred Beef × Dairy Cattle Does Not Impact Shape, Eating Quality, or Color of Strip Loin Steaks

Phenotypic expression of dairy influence often carries negative implications in beef production; thus, considerable variation in expression of beef- versus dairy-type might adversely affect value of crossbred beef × dairy cattle. This study evaluated effects of phenotype in crossbred beef × dairy cattle, specifically that associated with beef- versus dairy-type, on meat quality. Effects were blocked within commercial feedlot pens because cattle within a pen were contemporaries for sex, age, management, and source. On their harvest date, 592 Angus or [Simmental × Angus] × Holstein cattle from 9 pens were assessed by 3 expert evaluators. Scores for muscling and frame size were used to categorize and subset cattle in a pen into 4 phenotype groups: (1) fully dairy-type, (2) partially dairy-type, (3) partially beef-type, and (4) fully beef-type. Strip loin steaks were obtained from selected cattle (n = 82 to 84 per group) and evaluated for descriptive sensory attributes, shear force, pH, color at retail display, steak dimensions, muscle fiber type, and fatty acid composition. Data were tested for fixed effects of phenotype group with random effects of pen. Despite distinct expression of visual beef- versus dairy-type among cattle sampled, phenotype groups were largely not different (P &amp;gt; 0.05) in shape, sensory attributes, color, or biochemical properties of strip loin steaks. Other body regions, separate from the loin, were likely responsible for differences in live animal muscling. Additional research is needed on effects of sire breed, individual sire, and management strategies on meat quality in beef × dairy crossbreds. Complementarity of beef breeds and sires to produce more profitable beef-type cattle from the beef × dairy mating system should not be expected to negatively influence meat quality. Marketing programs rooted in production of consistent and premium products may benefit from including beef from beef × dairy crossbreds.

Performance of WLAN in Downlink MU-MIMO Channel with the Least Cost in Terms of Increased Delay

To improve the performance of IEEE 802.11 wireless local area (WLAN) networks, different frame-aggregation algorithms are proposed by IEEE 802.11n/ac standards to improve the throughput performance of WLANs. However, this improvement will also have a related cost in terms of increasing delay. The traffic load generated by mixed types of applications in current modern networks demands different network performance requirements in terms of maintaining some form of an optimal trade-off between maximizing throughput and minimizing delay. However, the majority of existing researchers have only attempted to optimize either one (to maximize throughput or minimize the delay). Both the performance of throughput and delay can be affected by several factors such as a heterogeneous traffic pattern, target aggregate frame size, channel condition, competing stations, etc. However, under the effect of uncertain conditions of heterogeneous traffic patterns and channel conditions in a network, determining the optimal target aggregate frame size is a significant approach that can be controlled to manage both throughput and delay. The main contribution of this study was to propose an adaptive aggregation algorithm that allows an adaptive optimal trade-off between maximizing system throughput and minimizing system delay in the WLAN downlink MU-MIMO channel. The proposed approach adopted different aggregation policies to adaptively select the optimal aggregation policy that allowed for achieving maximum system throughput by minimizing delay. Both queue delay and transmission delay, which have a significant impact when frame-aggregation algorithms are adopted, were considered. Different test case scenarios were considered such as channel error, traffic pattern, and number of competing stations. Through system-level simulation, the performance of the proposed approach was validated over the FIFO aggregation algorithm and earlier adaptive aggregation approaches, which only focused on achieving maximum throughput at the expense of delay. The performance of the proposed approach was evaluated under the effects of heterogenous traffic patterns for VoIP and video traffic applications, channel conditions, and number of STAs for WLAN downlink MU-MIMO channels.

High speed two-photon laser scanning stereomicroscopy for three-dimension tracking multiple particles simultaneously in three-dimension.

In this paper, we will describe a video rate two-photon laser scanning stereomicroscopy for imaging-based three-dimensional particle tracking. Using a resonant galvanometer, we have now achieved 30 volumes per second (frame size 512 × 512) in volumetric imaging. Owing to the pulse multiplexing and demultiplexing techniques, the system does not suffer the speed loss for taking two parallax views of a volume. The switching time between left and right views is reduced to several nanoseconds. The extremely fast view switching and high volumetric imaging speed allow us to track fast transport processes of nanoparticles in deep light-scattering media. For instance, in 1%-intralipid solution and fibrillar scaffolds, the tracking penetration depth can be around 400 μm.

Short Frame Transmission at Very Low SNR by Associating CCSK Modulation With NB-Code

In this paper, we present a frame structure that can be viewed as a preamble for the detection and synchronization process (leading to low cost receiver) and as an encoded codeword carrying the transmitted message (leading to reliable transmission). This duality facilitates an ALOHA protocol avoiding preamble overhead. The frame structure, named Quasi-Cyclic Short Packet (QCSP), is based on the association of a Cyclic Code Shift Keying (CCSK) modulation and a non-binary error control code. The detection/correction algorithm of the QCSP system is presented, its performance is theoretically derived and discussed for different parameters. A QCSP frame can be transmitted and received correctly with an error probability of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathbf {10^{-4}}$ </tex-math></inline-formula> , distanced by <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1.2</b> dB from Polyanskiy’s bound (an estimated Shannon’s limit for small packet size) at −11 dB of SNR for a payload of 360 bits. Compared to a classical preamble-based frame using a modern binary error control code, the size of a QCSP frame is reduced by 23%. Moreover, detecting a QCSP frame requires a lower complexity than detecting a longer classical preamble.

Image Stitching Technology in 3D Film Production Based on Digital Image Processing

Image stitching in 3D film production synchronizes multiple independent frame images as a single high-resolution output. The stitching process is common for displaying a large coverage or shot in a single frame containing distinct images. For improving the realistic accuracy of 3D film images, this manuscript introduces an Itinerant Pixel-Matching-based Stitching Process (IPMSP). The proposed stitching process relies on cross-sectional pixels identified in merging two or more images. Based on the linear cumulative distribution in augmenting images, the homogeneity feature is identified. If the homogeneity is high then the stitching for linear pixels occurs increasing the frame resolution and size. The cumulative distribution is determined using the recurrent neural network in verifying homogeneity and contrast. For the contrast pixels identified, the cross-sectional matching is performed for substituting similar pixels in the missing sections. The process is repeated until the stitching region is the same based on the homogeneity feature and increased dimensions. Therefore, this process is capable of improving accuracy, precision, substitution, and reducing errors and complexity.

Artificial intelligence – Based video traffic policing for next generation networks

The constant increase in users’ bandwidth needs, through a large variety of multimedia applications, creates the need for highly effective network traffic control. This need is imperative in wireless networks, where the available bandwidth is limited, but is very important for wired networks as well. In this work we focus on the problem of policing video traffic from sources encoded with H.264 and H.265, given that these are the major state-of-the-art standards currently in the market. Building on work that has shown that classic traffic policing schemes can lead to unnecessarily strict policing for conforming video sources, we propose the use of Artificial Intelligence (AI) – based traffic policing schemes for video traffic. We conduct a performance evaluation of several AI – based schemes with the classic token bucket and we show that our proposed Frame Size Predictor and Policer scheme improve the performance of the classic token bucket by around 90% for conforming users, while providing only slightly worse policing results for non-conforming users.

Automatic Extraction of Power Lines from Aerial Images of Unmanned Aerial Vehicles.

Automatic power line extraction from aerial images of unmanned aerial vehicles is one of the key technologies of power line inspection. However, the faint power line targets and complex image backgrounds make the extraction of power lines a greater challenge. In this paper, a new power line extraction method is proposed, which has two innovative points. Innovation point one, based on the introduction of the Mask RCNN network algorithm, proposes a block extraction strategy to realize the preliminary extraction of power lines with the idea of “part first and then the whole”. This strategy globally reduces the anchor frame size, increases the proportion of power lines in the feature map, and reduces the accuracy degradation caused by the original negative anchor frames being misclassified as positive anchor frames. Innovation point two, the proposed connected domain group fitting algorithm solves the problem of broken and mis-extracted power lines even after the initial extraction and solves the problem of incomplete extraction of power lines by background texture interference. Through experiments on 60 images covering different complex image backgrounds, the performance of the proposed method far exceeds that of commonly used methods such as LSD, Yolact++, and Mask RCNN. DSCPL, TPR, precision, and accuracy are as high as 73.95, 81.75, 69.28, and 99.15, respectively, while FDR is only 30.72. The experimental results show that the proposed algorithm has good performance and can accomplish the task of power line extraction under complex image backgrounds. The algorithm in this paper solves the main problems of power line extraction and proves the feasibility of the algorithm in other scenarios. In the future, the dataset will be expanded to improve the performance of the algorithm in different scenarios.