The Oriented FAST and Rotated BRIEF (orb) algorithm plays a crucial role in rapidly extracting image keypoints. However, in the domain of high-frame-rate real-time applications, the algorithm faces challenges of the s...
详细信息
The Oriented FAST and Rotated BRIEF (orb) algorithm plays a crucial role in rapidly extracting image keypoints. However, in the domain of high-frame-rate real-time applications, the algorithm faces challenges of the speed and computational efficiency with the increase in both the size and quantity of images. To address this issue, an orb algorithm accelerator based on a computing-in-memory (CIM) circuit is firstly proposed in this paper, which replaces the iterative calculations in traditional methods with one-step parallel analog computation. The proposed accelerator improves algorithm computational efficiency through CIM technology and enhances algorithm speed through parallel computation. Simulation demonstrate that the proposed method exhibits an average processing speed 22 x faster than traditional methods and obtains more uniform corners distribution in large-scale images.
In recent years, with the continuous development of society, people's exploration in the field of machine vision images has gradually increased the demand for machine vision and digital image processing. Only the ...
详细信息
ISBN:
(纸本)9783031243660;9783031243677
In recent years, with the continuous development of society, people's exploration in the field of machine vision images has gradually increased the demand for machine vision and digital image processing. Only the consistency comparison research of machine vision images based on the improved orb algorithm, in order to obtain more complete image information. Based on the improved orb algorithm, this paper conducts research by improving the consistency comparison of machine vision images, and meets the requirements of accuracy and precision for the obtained machine vision images and so on. This paper briefly introduces the technology and development trend of machine vision image consistency comparison, studies the machine vision image consistency comparison, and through a series of experiments to prove that the machine vision image consistency comparison based on the improved orb algorithm is effective in To a certain extent, it has certain feasibility in terms of precision and accuracy. Analysis and comparison based on different image stitching methods were carried out. The final results of the research show that the accuracy of the five-consistency comparison of machine vision images is 98.7% when the distance of image five is 83.4 km. Experimental data show that the accuracy of machine vision image consistency comparison has always been maintained at a stable level, that is, 97%. It shows that the accuracy of machine vision image consistency comparison does not decrease with the increase of distance.
UAVs are flexible in action, changeable in shooting angles, and complex and changeable in the shooting environment. Most of the existing stitching algorithms are suitable for images collected by UAVs in static environ...
详细信息
UAVs are flexible in action, changeable in shooting angles, and complex and changeable in the shooting environment. Most of the existing stitching algorithms are suitable for images collected by UAVs in static environments, but the images are in fact being captured dynamically, especially in low-altitude flights. Considering that the great changes of the object position may cause the low-altitude aerial images to be affected by the moving foreground during stitching, so as to result in quality problems, such as splicing misalignment and tearing, a UAV aerial image stitching algorithm is proposed based on semantic segmentation and orb. In the image registration, the algorithm introduces a semantic segmentation network to separate the foreground and background of the image and obtains the foreground semantic information. At the same time, it uses the quadtree decomposition idea and the classical orb algorithm to extract feature points. By comparing the feature point information with the foreground semantic information, the foreground feature points can be deleted to realize feature point matching. Based on the accurate image registration, the image stitching and fusion will be achieved by the homography matrix and the weighted fusion algorithm. The proposed algorithm not only preserves the details of the original image, but also improves the four objective data points of information entropy, average gradient, peak signal-to-noise ratio and root mean square error. It can solve the problem of splicing misalignment tearing during background stitching caused by dynamic foreground and improves the stitching quality of UAV low-altitude aerial images.
Feature points obtained using traditional orb methods often exhibit redundancy, uneven distribution, and lack scale invariance. This study enhances the traditional orb algorithm by presenting an optimal technique for ...
详细信息
Feature points obtained using traditional orb methods often exhibit redundancy, uneven distribution, and lack scale invariance. This study enhances the traditional orb algorithm by presenting an optimal technique for extracting feature points, thereby overcoming these challenges. Initially, the image is partitioned into several areas. The determination of the quantity of feature points to be extracted from each region takes into account both the overall number of feature points and the number of divisions that the image undergoes. This method tackles concerns related to the overlap and redundancy of feature points in the extraction process. To counteract the non-scale invariance issue in feature points obtained via the orb method, a Gaussian pyramid is employed, and feature points are extracted at each level. Experimental findings demonstrate that our method successfully extracts feature points with greater uniformity and rationality, while preserving image matching accuracy. Specifically, our technique outperforms the traditional orb algorithm by approximately 4% and the SURF algorithm by 2% in terms of matching performance. Additionally, the processing time of our proposed algorithm is three times faster than that of the SURF algorithm and twelve times faster than the SIFT algorithm.
Considering the scale transformation and the influence of rotation angle among image feature point extraction process, to improve the low efficiency and poor stability, an improved SRBICP algorithm which combined orb ...
详细信息
ISBN:
(纸本)9781450372589
Considering the scale transformation and the influence of rotation angle among image feature point extraction process, to improve the low efficiency and poor stability, an improved SRBICP algorithm which combined orb and ICP algorithm is proposed and applied to register the point cloud, Based on this method, the boundary of the scale matrix is considered and the rotation angle constraint is added. The point cloud registration model is optimized by considering some factors such as matrix, dynamic iteration coefficient, and annealing coefficient. Benefit from the Ubuntu operating system which is based on the open source GNU/Linux system, the simulation experiments are carried out. The experimental results show that the improved algorithm performs better on the registration accuracy as about 40% when being compared with the traditional ICP algorithm, and the registration speed is increased as about 60%.
The traditional speeded-up robust features (SURF) algorithm has certain stability in scale, rotation, illumination and other changes. However, this algorithm has problems such as large amount of computation, low match...
详细信息
The traditional speeded-up robust features (SURF) algorithm has certain stability in scale, rotation, illumination and other changes. However, this algorithm has problems such as large amount of computation, low matching accuracy, and time-consuming in feature extraction and feature matching. An improved adaptive orb-SURF image matching algorithm is proposed in this paper. Fusing edge features and using improved Oriented FAST and Rotated BRIEF (orb) algorithms are used by this algorithm to extract image feature points. Moreover, SURF descriptors are used by feature points for feature description. Then an improved fast library for approximate nearest neighbors (FLANN) algorithm was used for adaptive feature matching. The random sample consensus (RANSAC) algorithm is used to eliminate the false matching point pairs after the selected points to be matched. Finally, the excellent matching point pairs reserved by the adaptive FLANN algorithm are combined with the excellent matching point pairs reserved by the RANSAC algorithm to complete the matching. The experimental results show that the average accuracy of the improved algorithm can reach more than 98%, which is about 6% higher than the original SURF algorithm. And the average matching time is 1.2S, which is about 25% lower than the original SURF algorithm. It is worth mentioning that the problem that the original SURF algorithm cannot predict the number of feature points through Hessian threshold is solved by this algorithm. Moreover, compared with the SuperPoint based deep learning image matching algorithm, the image matching time of this algorithm is reduced by 80%.
This paper proposes an image matching algorithm (L-SURB algorithm) based on the SURF algorithm and the orb algorithm. The process of algorithm can be divided into four steps. Firstly, the image is enhanced by Laplacia...
详细信息
ISBN:
(纸本)9781538695944
This paper proposes an image matching algorithm (L-SURB algorithm) based on the SURF algorithm and the orb algorithm. The process of algorithm can be divided into four steps. Firstly, the image is enhanced by Laplacian operator. Secondly SURF detector is used to detect feature points. Thirdly, the orb descriptor is used to describe the feature points to generate a rotation invariant binary descriptor. Finally, the rough matching of the feature points is completed by Hamming distance and the exact matching is realized by Lowe's algorithm. The results of experiment show that L-SURB algorithm effectively solves the problem that orb algorithm is sensitive to image brightness and lacking in scale invariance, which greatly improves the matching accuracy. At the same time, the matching speed of L-SURB algorithm is increased by 81.5% compared with SURF algorithm.
To solve the problem of slow image feature extraction and poor matching accuracy in mobile augmented reality (MAR) scene recognition, this paper proposes an improved natural feature recognition algorithm based on SURF...
详细信息
ISBN:
(纸本)9781538674130
To solve the problem of slow image feature extraction and poor matching accuracy in mobile augmented reality (MAR) scene recognition, this paper proposes an improved natural feature recognition algorithm based on SURF ( Speeded Up Robust Features) and orb (Oriented FAST and Rotated BRIEF) . Firstly, the image is preprocessed, including Gaussian smoothing, grayscale and histogram equalization to reduce the impacts of noise on image feature extraction;the image can be normalized through extracting useful information of image features, and then the image center will be selected as the feature point. Secondly, the SURF and orb algorithm are respectively used to describe the image feature points to determine the orientation of the image feature points so that the image could have the rotation invariance. Finally, the K-Nearest Neighbors (KNN) algorithm is used to select the SURF space and the orb spatial neighboring image, respectively, and the image weight, ie. the weighted KNN algorithm, is given to form a new image set, and the image with the smallest weight is selected as the matching image. Experimental results show that when the feature extraction time and matching time are less than 3 ms on Ordinary laptop, meanwhile the image matching accuracy is as high as 92.5%, the computing speed and matching accuracy are better than traditional algorithms. Therefore, the natural feature recognition of the image can be realized in real time and accurately.
The image registration technique is very important in the image processing and computer vision field, and widely used in object recognition, image analysis and target detection. The paper presents a markless augmented...
详细信息
ISBN:
(纸本)9781479921867
The image registration technique is very important in the image processing and computer vision field, and widely used in object recognition, image analysis and target detection. The paper presents a markless augmented reality system for natural scene restoration through which we can see the original appearance of the destructed scene. Based on the system, we put forward a mark-less image registration algorithm based on orb algorithm. The method uses the orb algorithm to extract the image feature points and feature descriptors, then eliminates the error matching by using random sampling consistency algorithm, finally obtains the registration information by the improved homography matrix. Experiments show that this method has a better real-time performance in image registration, reliability and positioning.
As people's daily behavioral activities become more data-based, how to protect personal information security is a crucial consideration for the whole society. Finger vein recognition is becoming an essential means...
详细信息
As people's daily behavioral activities become more data-based, how to protect personal information security is a crucial consideration for the whole society. Finger vein recognition is becoming an essential means of identification because of its uniqueness, live detection, security, and many other advantages. Although deep learning can make finger vein recognition have an excellent effect. However, the number of samples needed to build a deep network model is too large, and the current authoritative finger vein database cannot reach the minimum number of samples required. The emergence of Muti-Grained Cascade Forest provides a solution to the problem of insufficient sample data and long training time, which can give a new research avenue in feature extraction. In order to obtain higher accuracy, the deep forest algorithm is introduced in this paper to process the finger vein images. Firstly, the image data in the finger vein image database is pre-processed to prepare for the subsequent feature extraction and matching. Then, the deep forest algorithm is used to find the feature points, and the orb algorithm is used to match the features to obtain the angular information of each matched pair, and the final identity is determined according to the sparse distribution of angles. The accuracy of finger vein recognition based on the deep forest algorithm is 98.40%. By comparing with other machine learning methods for finger vein recognition, the proposed method has a higher accuracy rate.
暂无评论