SURF Feature Points Research Articles

Composite Fixed-Length Ordered Features with Index-of-Max Transformation for High-Performing and Secure Palmprint Template Protection

Palmprint recognition has attracted considerable attention due to its advantages over other biometric modalities such as fingerprints, in that it is larger in area, richer in information and able to work at a distance. However, the issue of palmprint privacy and security (especially palmprint template protection) remains under-studied. Among the very few research works, most of them only use orientational features of the palmprint with transformation processing, yielding unsatisfactory recognition and protection performance. Thus, this research work proposes a palmprint feature extraction method for palmprint template protection that is fixed-length and ordered in nature, by fusing point features and orientational features. Firstly, dual orientations are extracted and encoded with more accuracy based on the Modified Finite Radon Transform (MFRAT). Then, SURF feature points are extracted and converted to be fixed-length and ordered features. Finally, Composite Fixed-Length Ordered Features that fuse up the dual orientations and SURF points are transformed using the irreversible transformation of Index-of-Max (IOM) to generate the revocable palmprint templates. Experiments show that the matching accuracy of the proposed method of fixed-length and ordered point features are superior to all other feature extraction methods on the PolyU and CASIA datasets. It is also demonstrated that the EERs before and after IOM transformation are better than all other representative template protection methods. A thorough security and privacy analysis including brute-force attack, false accept attack, birthday attack, attack via record multiplicity, irreversibility, unlinkability and revocability is also given, which proves that our proposed method has both high performance and security.

Investigation of Medical Image Technology Based on Big Data Neuroscience in Exercise Rehabilitation.

The purpose of this article is to combine the functional information of CT images with the anatomical and soft tissue information of MRI through image fusion technology, providing more detailed information for rehabilitation treatment and thus providing a scientific basis for clinical applications and better training plans. In this paper, functional brain imaging technology combining CT (computed tomography) and MRI (magnetic resonance imaging) was used for image fusion, and SURF (accelerated robust feature) feature points of images were extracted. In this study, 40 patients with mild and moderate closed traumatic brain injury admitted to the rehabilitation department of a rehabilitation center from 2018 to 2022 were selected as the research objects. Compared with using only CT images and MRI images for brain injury diagnosis, the fusion image had a higher detection rate of abnormal brain injury diagnosis, with a detection rate of 97.5%. When using fused images for the diagnosis of abnormal brain injury, the patient's exercise rehabilitation effect was better. CT and MRI image fusion technology had a high diagnostic accuracy for brain injury, which could timely guide doctors in determining exercise rehabilitation plans and help improve the effectiveness of patient exercise rehabilitation.

Directional region-based feature point matching algorithm based on SURF.

Feature point matching is one of the fundamental tasks in binocular vision. It directly affects the accuracy and quality of 3D reconstruction. This study proposes a directional region-based feature point matching algorithm based on the SURF algorithm to improve the accuracy of feature point matching. First, same-name points are selected as the matching reference points in the left and right images. Then, the SURF algorithm is used to extract feature points and construct the SURF feature point descriptors. During the matching process, the location relationship between the query feature point and the reference point in the left image is directed to determine the corresponding matching region in the right image. Then, the matching is completed within this region based on Euclidean distance. Finally, the grid-based motion statistics algorithm is used to eliminate mismatches. Experimental results show that the proposed algorithm can substantially improve the matching accuracy and the number of valid matched points, particularly in the presence of a large amount of noise and interference. It also exhibits good robustness and stability.

Research on image recognition and processing of motion targets of warehouse logistics robots

Abstract In developing robots for warehouse logistics, image recognition and processing for moving targets are the cornerstone of subsequent work. In this paper, the Meanshift algorithm is extended to continuous image sequences, and the Camshift algorithm for motion target tracking in a warehouse environment is proposed to obtain effective tracking of targets through the probability distribution when the color of continuous images changes dynamically. Based on target tracking, a feature-matching-based image recognition method is constructed. The scene image is first treated with improved Gamma correction for light equalization, and then image features are extracted using SURF feature points. Regarding running time, the feature matching method is, on average, 2.03 seconds faster than FLDA and 0.96 seconds faster than PCAFLDA under the same external conditions. By optimizing the computational structure, the feature-matching method can address the need for efficiency in warehouse logistics.

Automatic Measurement of Elongation at Break of Cable Sheath Based on Binocular Vision

<p indent=0mm>In order to solve the problem of accuracy and automation of manual measurement for elongation at break of cable sheath, an automatic measurement based on binocular vision is proposed. Firstly, after the sample tensile video is collected, location method based on information entropy is used on the first frame for the center coordinates of two circle markers, if markers are correctly identified, the center coordinate is used as the marked point, otherwise the point closest to the center of circle in SURF feature points with good stereo matching results is selected as the marked point. Secondly, the marked points tracking based on L-K optical flow is used in the sampled video of sheath stretching process, the determination of the rough frame position of sheath breaking is based on the calculated three-dimensional coordinates and acceleration, then two sampling intervals are backtracked for frame by frame detection. Finally, the breaking frame is located precisely then elongation at break is calculated. The experimental results in the field of quality inspection of cable sheath show that the measured root mean square error of elongation at break is 7.13% which is lower than that of manual measurement method (21.84%), compared with the true values measured by 240 fps high-speed camera, and the processing speed of the method reaches 107.8 fps. Propose method can obviously improve the detection accuracy, speed and automation in practical application.

Health information acquisition and position calculation of plug seedling in greenhouse seedling bed

The health information acquisition of plug seedlings plays an important role in automatic seedling transplanting. This study proposed the use of a mobile inspection and transplanting machine operation method on the seedbed. Compared with the operation of a fixed transplanter, this method can reduce the workload of tray transportation. When the performing mobile picks and supplements the seedlings, the plug seedling detection system needs to obtain a large-field seedbed image and extract and locate multiple plug trays. Then, the system evaluates the health of the plug seedlings. This paper studied and compared the stitching effect of large-field seedbed images based on block area matching, Harris corner detection, and SURF feature point detection. Based on the large field seedbed image, the concentric rectangular frame template matching algorithm was designed and developed. This algorithm can realize the precise positioning of multiple plug trays on the large field seedbed image. Thus, the health status of plug seedlings was further assessed according to the area of seedlings in each cell. A 200-cell tray seedbed was the research object, and a digital camera was used to capture the partitioned image of the seedbed in multiple sections. The image stitching algorithm was used to obtain the stitched image of the large-field seedbed. After image preprocessing, including grayscale conversion, binary segmentation, and Gaussian filtering, the concentric rectangular frame template matching algorithm was used to stitched images to achieve the scanning positioning of the multiple plug trays. Furthermore, based on the positioning of the plug tray, the plug tray area was divided, and then the health status of plug seedlings was evaluated. Experiments have proved that the accuracy rate of seedling health information detection is 98.7%. Such results provide basis for precision management of seedling replenishment operations on the seedling bed and meets the requirements of actual operations.

Vision Guiding System Based on Strong Positioning and Hand-eye Calibration

In order to solve the current localization algorithm big background interference affects calibration methods exist damage, physical contact with the workpiece material on the surface of the problem, hand-eye calibration was proposed based on strong orientation and three visual guidance system. First, deepen the Hessian matrix, using Gaussian filter, using the maximum inhibition precisely determine the feature points and direction, in order to construct surf feature points description, finish strong positioning operator based on surf, to remove the background and the ambient light interference, the purpose of accurate positioning center of the target with Angle. And then put forward three points calibration method, calculation scale rotation matrix and translation matrix parameters, complete the Robot world coordinates and binding mapping camera image coordinates. Finally, program implementation algorithm and system, the experiment results showed that compared with the current workpiece material transfer system, this system has higher goal orientation and calibration transfer success rate.

Near Duplicate Image Retrieval using Multilevel Local and Global Convolutional Neural Network Features

In this work, we present an approach based on multilevel local as well as global Convolutional Neural Network (CNN) feature matching to retrieve near duplicate images. CNN features are suitable for visual matching. The CNN features of entire image may not give accuracy in retrieval due to various image editing/capturing operations. Our retrieval task focuses on matching image pairs based on local and global levels. In local matching, an image is segmented into fixed size blocks followed by extracting patches by considering neighboring regions at different levels. Matching local image patches at different levels provides robustness to our retrieval model. In local patch extraction, we select blocks containing SURF feature points instead of selecting all blocks. CNN features are extracted and stored for each image patch and then followed by extraction of global CNN features. Finally, similarity between image pairs is computed by considering all extracted CNN features. Our similarity function is based on correlation and number of blocks found in matching. We implemented our proposed approach on benchmarking Holiday dataset. Retrieval results show remarkable improvement in mean average precision (mAP) on the dataset.

Image precise matching with illumination robust in vehicle visual navigation

In vehicle visual navigation, image matching algorithm is highly critical to positioning accuracy and processing efficiency. One single matching algorithm cannot satisfy all types of image features accurate acquisition, so Harris, SUSAN, FAST, SIFT, and SURF are respectively adopted to process various road images under normal lighting condition. During practical application, the appropriate algorithm can be selected based on detection rate and running time of the above algorithms. Aiming at the illumination change interference of the collected images in vehicle visual navigation, many traditional matching algorithms for illumination change are not optimal, so an image precise matching algorithm with illumination change robustness is proposed. Because image edges and detail information have lower sensitivity for illumination change, SURF feature points are optimized by image gradient based on the idea of Canny, and the bidirectional search is used to obtain precise matching points. The experimental results show that feature point detection of the algorithm remains good stability for illumination change in images, and the matching accuracy can reach more than 94%. The algorithm is not only robust to illumination change, but also ensures higher matching speed and meanwhile improves the matching accuracy significantly.

Distribution of Landscape Architecture Based on 3D Images and Virtual Reality Rationality Study

In order to ensure the rationality of landscape design, it is of great practical significance to simulate the planning effect. Aiming at the problems of traditional garden landscape distribution simulation technology, this paper designs a new garden landscape distribution rationality simulation technology based on three-dimensional image and virtual reality technology. Firstly, a rational analysis method for landscape garden distribution based on three-dimensional images is proposed. Based on the correlation between different coordinate systems under a specific angle of view, a camera pose matrix is constructed, and SURF feature point detection is performed on the landscape garden image and the above feature points are clustered and matched. Secondly, the three-dimensional image reconstruction of the landscape garden image is realized based on the registered image feature points. Based on the three-dimensional image reconstruction results and virtual reality technology, the three-dimensional image analysis of the rationality of landscape distribution is completed. Finally, the case analysis and performance test results illustrate the superiority of the realized landscape garden distribution rationality technology. This method can provide scientific reference and basis for the modern landscape gardening industry to adopt three-dimensional images and other technologies.

Extraction of Iris Crypt, Pigment Spot, and Wolfflin Nodule Biological Feature using Feature Point Selection Algorithms

The iris biometrics is an important biological feature of the human. The iris is the part of human eye. The human eye consists of many features. Iris is one of the unique features of human eye. In this paper we propose an algorithm to extract the features of iris. The existing algorithms are based on combined biological features of iris. We are going to introduce separate biological features and extract them one by one using suitable algorithms. The proposed method is used to extract the biological features of human iris. The proposed method uses crypts, pigment layers, and Wolfflin nodules features of iris. Each feature is extracted initially and suitable feature selection algorithm is identified. The manual cropping is initially applied in the eye image which extracts iris layer. The manual cropping is further applied on iris to locate the biological layers. Canny edge detection is applied on each iris feature. The FAST, SURF, MinEigen, BRISK, and MSER feature points are collected from each biological layer. The MinEigen extracts 218 feature points from the crypt layer. The BRISK extracts 161 and 89 feature points from the pigment and Wolfflin nodules. The proposed system can be used in iris recognition system all over the world for various authentication and security purposes. The individual feature extraction helps to make the recognition system more secure and accurate as compared to the existing systems which uses combined biological features. Thus, the proposed system is advantageous as compared to the existing systems and the efficiency will also be high if it is used for iris recognition.

MOSAICING FIDELITY GEOMETRICAL ASSESSMENT BASED ON SURF POINT CLASSIFICATION

Abstract. The recent development of automated UAV imaging applications for geomatics is leading to an unprecedented rapid growth in mosaic and DEM images. Newer, more advanced algorithms are being extensively studied to fulfill constantly increasing requirements. Sequences of georeferenced images that cannot be shot at once are merged with ideally no geometrical distortion to allow for 3D reconstruction and orthophotography generation. If newer robust mosaicing algorithms are being developed to withstand UAV tighter real-time constraints, very little attention has been given to objectively assess their capacity to coherently combine different image geometries. Changes in FOV and illumination, oblique scenes and dynamic sequences featuring movement and object occlusion are not yet fully handled by state-of-the-art algorithms. Despite the large panel of existing applications featuring different quality constraints, our work focuses on the geomatics context, where the main requirement is the respect of geometrical proportions among images. Geometrically coherent mosaics and 3D reconstructions can only result from fully static, high-altitude orthogonal views. GCPs or well known geographical landmarks are employed with RTK-GPS for scene georeferencing. The overall fidelity is computed as the global variance of the distances between SGPs. However, very few ways seem to exist to measure the content coherence of large geographical scenes, interiors or unreferenced scenes. Distances are mostly expressed in a local coordinate system and GCPs are often unavailable. The main contribution of this paper consists in the definition of a formal metric for measuring local coordinate geometrical fidelity of mosaics issued of UAV image sequences, in real operating conditions. A Mosaicing Fidelity Assessment (MSF) index is therefore computed for every couple of consecutive images of a sequence. The metric generates an index based on the distances of SURF feature points extracted in the images and compares them to estimate geometrical changes transferred to the mosaic. The solution can detect perspective inaccuracies caused by residual registration errors. It can fulfill the requirements of most of imaging and geomatics applications and can be executed as nearly real-time due to its low computational complexity.

Protagonist and deuteragonist-based video indexing and retrieval system for movie and video song sequences

Protagonist and deuteragonist are two main characters that plays leading role in Indian Hindi movie (IHM). Currently such information is attached using textual caption which is highly unreliable. The research presented in this paper is to automatically index and retrieve the content based on protagonist and deuteragonist from the large IHM and video song sequences (VSS). Video song sequence indexes are extracted using audio-based approach from IHM in module 1. These indexes are used as an input in module 2. Faces are identified from every VSS. Colour histogram and spatiogram descriptor are extracted from faces. Similarity between two faces is computed using Bhattacharyya coefficient. Similarity-based clustering technique is performed to obtain clusters of faces. Recognition of protagonist and deuteragonist is done using surf feature points. The experimental results are carried out using Indian Hindi movies of different genres.

Edge-Preserving Stereo Matching Using Minimum Spanning Tree

Despite that the accuracy and efficiency of stereo matching technology have significantly improved in the past decades, the issue of edge-blurring remains a challenge to most of the existing approaches. In this paper, we propose a minimum spanning tree (MST) based stereo matching method by using the image edge and segmentation optimization to preserve the image boundary. We first exploit a fast disparity range estimation method by combining the Surf and Akaze feature points to improve the computational efficiency. Second, we utilize the image edges and brightness information to generate a self-adaptive weight function, which is able to significantly improve the accuracy of MST aggregating in the regions of complex texture and boundaries with similar color distribution. Third, we employ the image segmentation to extract the invalid regions of the estimated disparity map, and propose a post-processing scheme to refine the disparity result. Finally, we run our method on several Middlebury and KITTI datasets. The comparison results between our method and other state-of-the-art approaches demonstrate that the proposed method has high accuracy for disparity computation and is especially robust to the edge-blurring.

Fast recognition algorithm for static traffic sign information

Abstract Aiming at the low recognition rate, low recognition efficiency, poor anti-interference and high missing detection rate of current traffic sign recognition methods, a fast recognition algorithm based on SURF for static traffic sign information of highway is proposed. The expansion of the digital morphological method is used to connect the cracks in the traffic sign. Traffic sign images are corroded according to the corrosion, and the connected areas are contracted or refined. Regions of interest are detected by region filling. According to the result of traffic sign image processing, the scale of traffic sign image is normalized by bilinear interpolation method, and the SURF feature points of traffic sign image are extracted. The FLANN algorithm is used to realize feature point matching, and the threshold is set to determine the best matching point. The matching result is output and the traffic sign information is recognized. Experimental results show that the algorithm has high recognition rate and recognition efficiency, strong anti-interference, and can control the rate of missing detection in a certain range.

Novel Registration and Fusion Algorithm for Multimodal Railway Images with Different Field of Views

Objects intruding high-speed railway clearance do great threat to running trains. In order to improve accuracy of railway intrusion detection, an automatic multimodal registration and fusion algorithm for infrared and visible images with different field of views is presented. The ratio of the nearest to next nearest distance, geometric, similar triangle, and RANSAC constraints are used to refine the matching SURF feature points successively. Correct matching points are accumulated with multiframe to overcome the insufficient matching points in single image pair. After being registered, an improved Contourlet transform fusion algorithm combined with total variation and local region energy is proposed. Inverse Contourlet transform to low frequency subband coefficient fused with total variation model and high frequency subband coefficients fused with local region energy is used to reconstruct the fused image. The comparison to other 4 popular fusion methods shows that our algorithm has the best comprehensive performance for multimodal railway image fusion.

Fast Visual Tracking Based on Convolutional Networks

Recently, an upsurge of deep learning has provided a new direction for the field of computer vision and visual tracking. However, expensive offline training time and the large number of images required by deep learning have greatly hindered progress. This paper aims to further improve the computational performance of CNT which is reported to deliver 5 fps performance in visual tracking, we propose a method called Fast-CNT which differs from CNT in three aspects: firstly, an adaptive k value (rather than a constant 100) is determined for an input video; secondly, background filters used in CNT are omitted in this work to save computation time without affecting performance; thirdly, SURF feature points are used in conjunction with the particle filter to address the drift problem in CNT. Extensive experimental results on land and undersea video sequences show that Fast-CNT outperforms CNT by 2~10 times in terms of computational efficiency.

Body Height Estimation System Based on Binocular Vision

In this paper, we propose a novel approach to estimate body height from video sequences based on binocular stereo vision. Firstly, we built a parallel binocular stereo vision device and detected the foreground by using Gaussian mixture model. After shadow elimination, we proposed the contour screening algorithm to obtain the human foreground and the top point in the foreground image. Then, we detected SURF feature points in the binocular images and screened them for 3 times to calculate the disparity of the head. After that, the height of human bodies can be estimated with the calibration parameters of binocular cameras. The experimental results demonstrate that the proposed method has higher measurement accuracy and spends less time which proves the effectiveness of the method.

Key Frame Extraction of Surveillance Video based on Moving Object Detection and Image Similarity

For the traditional method to extract the surveillance video key frame, there are problems of redundant information, substandard representative content and other issues. A key frame extraction method based on motion target detection and image similarity is proposed in this paper. This method first uses the ViBe algorithm fusing the inter-frame difference method to divide the original video into several segments containing the moving object. Then, the global similarity of the video frame is obtained by using the peak signal to noise ratio, the local similarity is obtained through the SURF feature point, and the comprehensive similarity of the video image is obtained by weighted fusion of them. Finally, the key frames are extracted from the critical video sequence by adaptive selection threshold. The experimental results show that the method can effectively extract the video key frame, reduce the redundant information of the video data, and express the main content of the video concisely. Moreover, the complexity of the algorithm is not high, so it is suitable for the key frame extraction of the surveillance video.

Study on the traditional pattern retrieval method of minorities in Gansu province

The traditional patterns of ethnic minorities in gansu province are ethnic arts with strong ethnic characteristics. It is the crystallization of the hard work and wisdom of minority nationalities in gansu province. Unique traditional patterns of ethnic minorities in Gansu province with rich ethnic folk arts, is the crystallization of geographical environment in Gansu minority diligence and wisdom. By using the Surf feature point identification algorithm, the feature point extractor in OpenCV is used to extract the feature points. And the feature points are applied to compare the pattern features to find patterns similar to the artistic features. The application of this method can quickly or efficiently extract pattern information in a database.