SURF Algorithm Research Articles

МЕТОД ДЕТЕКТУВАННЯ РУХУ ДЛЯ НЕСТАТИЧНИХ КАМЕР АБО ДЛЯ КАМЕР З ФУНКЦІЄЮ PTZ

Urgency of the research. Modern trends in the development of video surveillance systems focus on the creation of automatic functional complexes with intelligence. Detection of the areas of motion in a frame is a fundamental problem of these systems. Therefore, there is a need to create software that detects areas of motion in the frame.Target setting. By developing such systems, developers have to solve the contradiction between the computational complexity of processing algorithms, the quality of images created and the hardware capabilities of modern computing. The multidimensional nature of video information (coordinates, color) imposes significant limitations on the speed and quality of its processing. In addition, motion detection on mobile cameras is complicated due to constant background change.Actual scientific researches and issues analysis. The latest open access publications, including optical streaming technology and the SURF algorithm, were considered.Uninvestigated parts of general matters defining. Increasing the speed of existing methods of detecting traffic zones on the video of low quality.The research objective. Improvement of the method of detecting motion using non-static cameras or cameras with PTZ function in the direction of accelerating its operation.The statement of basic materials. Detection zone of motion in the scene comes with standard libraries openCV. To determine the zone of motion, a pair of sequential frames are used: the previous one - the object, the current one - the scene. Forming contours of objects occurs by constructing the corresponding histograms.Conclusions. The proposed method allows twice accelerating the process of detecting traffic zones by the given video of non-static cameras.

基于SURF算法的在线三维测量方法

基于SURF算法的在线三维测量方法

Bionic Vision-Based Intelligent Power Line Inspection System.

Detecting the threats of the external obstacles to the power lines can ensure the stability of the power system. Inspired by the attention mechanism and binocular vision of human visual system, an intelligent power line inspection system is presented in this paper. Human visual attention mechanism in this intelligent inspection system is used to detect and track power lines in image sequences according to the shape information of power lines, and the binocular visual model is used to calculate the 3D coordinate information of obstacles and power lines. In order to improve the real time and accuracy of the system, we propose a new matching strategy based on the traditional SURF algorithm. The experimental results show that the system is able to accurately locate the position of the obstacles around power lines automatically, and the designed power line inspection system is effective in complex backgrounds, and there are no missing detection instances under different conditions.

一种改进的SURF彩色遥感图像配准算法

一种改进的SURF彩色遥感图像配准算法

영상의 방향성을 이용한 ROI 기반 실시간 파노라마 영상 정합

본 논문은 임베디드 환경에서 실시간으로 동작하기 위해 입력 영상의 방향성을 판단하여 ROI를 지정하고 SURF, PROSAC 알고리즘을 적용하는 파노라마 영상 정합 방법을 제안한다. 기존의 파노라마 정합 알고리즘은 SURF 알고리즘이나 SIFT 알고리즘과 같이 많은 연산 데이터가 사용되는 알고리즘을 영상 전체에 적용하여 연산하기 때문에 불필요한 연산 영역을 포함하고 있다. 따라서 본 논문은 입력되는 영상의 방향을 판단한 결과로 영상의 ROI를 지정하여 불필요한 연산을 줄이는 방법을 제안한다. 영상의 방향을 판단하는데 사용하는 센서는 자이로 센서와 가속도 센서를 사용한다. 임베디드 환경에서 자이로 센서와 가속도 센서의 상보 필터를 통한 값 보정은 영상 정합 알고리즘의 처리 시간에 영향을 주지 않기 때문에 실시간 동작이 가능하다. In this paper, we proposed an implementation of panoramic image stitching that operates in real time at the embedded environment by applying ROI based PROSAC algorithm using the directionality of the image. The conventional panoramic image stitching applies SURF or SIFT algorithm which contains unnecessary computation and a lots of data to detect feature points. In this paper, we use the direction of the input image and we proposed the method of reducing the unnecessary computation by using ROI. We use a gyro sensor and an acceleration sensor. Output data from gyro and acceleration sensors can be calibrated by complementary filter. The calibration does not affect the operating time of the proposed image stitching algorithm in embedded environment. Therefore, it is possible to operate in real-time.

Classifying Bacterial Species using Computer Vision and Machine Learning

study embodied in this paper, aims at making use of machine learning and computer vision algorithms in order to reliably identify the species of bacteria, from their microscopic images. The study has taken into consideration three of the most commonly occurring species of bacteria that are clinically important. The work shown further in this study can be extended to a larger number of bacteria species. The study makes use of the Speeded Up Robust Features or SURF algorithm for detecting image keypoints. The artificial intelligence classifier makes use of these keypoint vectors as its input variable. It is noteworthy that the images used are taken at the gram staining stage of bacterial identification, in order to minimize any biases in the dataset owing to a variance in staining technique, although the feature detector invariably grayscales the image prior to keypoint computation. The paper follows the study from the conception of the idea, to the formulating of the algorithm, the training and testing of the same.

A machine vision based autonomous navigation system for Lunar rover: the model and key technique

Purpose This study aims to find a feasible precise navigation model for the planed Lunar rover. Autonomous navigation is one of the most important missions in the Chinese Lunar exploration project. Machine vision is expected to be a promising option for this mission because of the dramatic development of an image processing technique. However, existing attempts are often subject to low accuracy and errors accumulation. Design/methodology/approach In this paper, a novel autonomous navigation model was developed, based on the rigid geometric and photogrammetric theory, including stereo perception, relative positioning and absolute adjustment. The first step was planned to detect accurate three-dimensional (3D) surroundings around the rover by matching stereo-paired images; the second was used to decide the local location and orientation changes of the rover by matching adjacent images; and the third was adopted to find the rover’s location in the whole scene by matching ground image with satellite image. Among them, the SURF algorithm that had been commonly believed as the best algorithm for matching images was adopted to find matched images. Findings Experiments indicated that the accurate 3D scene, relative positioning and absolute adjustment were easily generated and illustrated with the matching results. More importantly, the proposed algorithm is able to match images with great differences in illumination, scale and observation angle. All experiments and findings in this study proved that the proposed method could be an alternative navigation model for the planed Lunar rover. Originality/value With the matching results, an accurate 3D scene, relative positioning and absolute adjustment of rover can be easily generated. The whole test proves that the proposed method could be a feasible navigation model for the planed Lunar rover.

The Integration of the Image Sensor with a 3-DOF Pneumatic Parallel Manipulator.

The study aims to integrate the image sensor for a three-axial pneumatic parallel manipulator which can pick and place objects automatically by the feature information of the image processed through the SURF algorithm. The SURF algorithm is adopted for defining and matching the features of a target object and an object database. In order to accurately mark the center of target and strengthen the feature matching results, the random sample and consensus method (RANSAC) is utilized. The ASUS Xtion Pro Live depth camera which can directly estimate the 3-D location of the target point is used in this study. A set of coordinate estimation calibrations is developed for enhancing the accuracy of target location estimation. This study also presents hand gesture recognition exploiting skin detection and noise elimination to determine the active finger count for input signals of the parallel manipulator. The end-effector of the parallel manipulator can be manipulated to the desired poses according to the measured finger count. Finally, the proposed methods are successfully to achieve the feature recognition and pick and place of the target object.

An enhanced SURF algorithm based on new interest point detection procedure and fast computation technique

In this paper, we propose an enhanced Speeded Up Robust Features (eSURF) algorithm to save memory and increase the operating speed. From analysis and observation of the conventional SURF algorithm, we show that a large amount of memory is inefficiently used to detect interest points and considerable operations are repeatedly performed when generating the descriptors of interest points. In the proposed algorithm, the scale-space representation (SSR) step and location (LOC) step are unified based on an efficient memory allocation technique to remove unnecessary memory. In addition, operations for Haar wavelet responses (HWRs) in horizontal and vertical directions, which occupy a major portion of computational loads, are performed by using a fast computation technique in which redundant calculations and repeated memory accesses are efficiently eliminated. Simulation results demonstrate that the proposed eSURF algorithm achieves a time savings of approximately 30 % and a memory savings of approximately 35.7 %, while the feature extraction performance of the proposed eSURF algorithm is exactly identical to that of the conventional SURF algorithm.

Mobile Robot Control System Based on a Modified 3D-Pointcloud Algorithm

The control system of mobile robot (functioning in buildings and structures), based on the methods of trajectories construction using a depth gauge sensor Kinect. The general approach of the work is based on the sequential execution of three phases: obstacle detection, mapping and construction barriers trajectories. Recognition guidelines robot is implemented using an SURF algorithm and FLANN library. Their combined use provides effective recognition obstacle in the near zone of the mobile robot to the required accuracy in real time. Mapping obstacles (concave shells) and the estimate of the distance to the border barriers is based on getting the so-called acceptable points. For this purpose, using the algorithm implemented Point-cloud detection of different types of obstacles in the close range of the robot. Synthesized a modified 3D-pointcloud algorithm which provides a solution to the problem of the robot move without colliding with obstacles. These algorithms can be applied to any stereo cameras and rangefinders that return colored frames and image depth, as described algorithms use data on the distance of objects from the sensor and are resistant to the visible spectrum of light (can operate in total darkness, without requiring a backlight). Built three-level robot control system in buildings and structures on the basis of a modified algorithm of 3D-pointcloud. The results of experiments confirming the effectiveness of the above approach.

Applying Artificial Bee Colony Algorithm for feature optimization in UAV Navigation

In this paper, the bee colony algorithm is used for optimizing features which have been extracted from UAV’s digital camera by SURF algorithm. To do that, the local map is stored in UAV memory before flight. During flight, images will be captured by a digital camera and the features in successive image will be extracted using SURF algorithm because SURF algorithm is considered highly insensitive to environmental light, scale changes and noise. Then, the extracted features will be optimized using the bee colony algorithm and will be compared with the original map features to find the location and direction of UAV. Simulations show that proposed algorithm has good precision and is robust to scale changes, light intensity variations, and noise.

Video Seamless Splicing Method Based on SURF Algorithm and Harris Corner Points Detection

In order to improve video image registration accuracy during splicing, a video image registration algorithm based on corner point detection and feature point integration is proposed in the article. Firstly, Harris algorithm is adopted to detect image corner points and fetch surf feature points of the image as well as integrate the corner points and feature points and eliminate the repeated points to generate feature vector; secondly, neural network is used to learn the image vector and establish image registration model, and the particle swarm algorithm is used to optimize the parameters of neural network; finally, the specific image is adopted for simulation comparison experiment. The result shows: compared with other image registration algorithms, the algorithm proposed in the article can improve image registration accuracy and efficiency and has relatively strong robustness.

Image Segmentation Method Based on SURF Algorithm and Harris Corner Points Detection

Image Segmentation Method Based on SURF Algorithm and Harris Corner Points Detection

Detecting Android malware phishing login interface based on SURF algorithm

Detecting Android malware phishing login interface based on SURF algorithm

Keypoint Extraction Using SURF Algorithm for CMFD

Copy-move forgery is a specific type of image tampering where a part of the image is copied and pasted on another part generally to conceal unwanted portions of the image. Hence, the goal in detection of copy-move forgeries is to detect image areas that are same or extremely similar. In this paper we propose a new approach which can efficiently identify the copy move forged areas. For achieving the copied image area first segment the image that is the test image is segmented into different independent patches. These patches are evaluated based on their matching and then the copied images are identified. For patch matching a two level of matching is carried out. In the first stage suspicious pairs of matches is obtained and affine transform matrix is roughly estimated. In-order to confirm the existence of copy move forgery a second level of patch matching with respect to Expectation Maximization algorithm is used. Thus the copy move forged areas are identified.

ROI 기반 실시간 이미지 정합 알고리즘 구현

본 논문은 임베디드 환경에서 실시간으로 동작하기 위해 이미지에 ROI를 지정하고 PROSAC 알고리즘을 적용하여 구현한 파노라마 영상 정합을 제안한다. 기존의 방식은 SURF 알고리즘이나 SIFT 알고리즘과 같이 복잡한 연산과 많은 연산 데이터의 알고리즘을 화면 전체에 적용하여 탐색한다. 또한 outlier 제거 알고리즘으로 RANSAC을 적용하여 알고리즘이 가진 무작위성으로 추가적인 검증 시간을 필요로 한다. 본 논문은 파노라마 영상의 특성을 고려하여 ROI를 설정함으로써 불필요한 연산량을 줄이고 outlier 제거 알고리즘을 검증 시간을 줄인 PROSAC 알고리즘으로 채택하여 처리 속도를 개선하였다. 비교 실험은 ARM Cortex-A15가 내장된 ODROID-XU 환경에서 진행 하였다. 제안하는 방식은 기존의 방식보다 처리 시간이 약 54% 개선되었다. This paper proposes a panoramic image stitching that operates in real time at the embedded environment by applying ROI and PROSAC algorithm. The conventional panoramic image stitching applies SURF or SIFT algorithm which contains complicated operations and a lots of data, at the overall image to detect feature points. Also it applies RANSAC algorithm to remove outliers, so that an additional verification time is required due to its randomness. In this paper, unnecessary data are eliminated by setting ROI based on the characteristics of panorama images, and PROSAC algorithm is applied for removing outliers to reduce verification time. The proposed method was implemented on the ORDROID-XU board with ARM Cortex-A15. The result shows an improvement of about 54% in the processing time compared to the conventional method.

Методики сопоставления особых точек в задаче визуальной навигации БПЛА

The paper is devoted to development and experimental comparison of techniques feature points matching on the images — images of the earth's surface with cameras mounted on un-manned aerial vehicles (UAVs) and artificial earth satellite. The main feature of the problem is that one of the images (satellite image) is divided into fragments. The developed techniques are part of a complex of algorithms for determining the position and orientation of the UAV using the methods and algorithms of machine vision. A description of the flight simulation technology and solving positioning tasks are described. Feature points on the image are extracted using an SURF algorithm. Also the approach to matching based on the partition of feature points’ set into two subsets depending on the sign of the Laplacian are stidied. Methods of increasing the performance of matching points are offered.

A Method Using Auxiliary Direction to Improve SURF Recall

Classic SURF algorithm may lead to matching failure, low recall because of incorrect main direction when constructing feature points describing operator. To solve it, A Method using auxiliary direction to improve SURF recall is put forward. The improved algorithm first select out auxiliary direction which is similar to main direction in characteristics, then generate new operator for describing the auxiliary direction characteristic. When matching, the improved algorithm adopts stricter nearest neighbor proportion inhibition. Experimental results show that feature point recall increase about 6% compared with the classical SURF while maintaining the precision.

Purwarupa Sistem Penggabungan Foto Udara Pada UAV Menggunakan Algoritma Surf (Speeded-Up Robust Features)

Abstrak Purwarupa penggabungan foto udara pada UAV menggunakan algoritma SURF merupakan suatu sistem yang dirancang untuk melakukan penggabungan citra. Citra tersebut adalah citra yang dihasilkan fixed-wings UAV. Keluaran dari sistem ini berupa tampilan citra dengan objek yang lebih luas. Sistem ini dirancang untuk dapat melakukan penggabungan foto udara dengan menggunakan algoritma SURF, FLANN, RANSAC, dan warpPerspective. Algoritma SURF digunakan sebagai detektor keypoint dari masing-masing input foto. Metode FLANN untuk melakukan pencocokan keypoint yang ditemukan. RANSAC digunakan untuk pencarian matrix homography. Metode warpPerspective digunakan untuk penggabungan kedua input yang memiliki kecocokan keypoint. Pengujian terdiri dari beberapa jenis variasi antara lain variasi minimal perpotongan, variasi skala dan variasi rotasi. Variasi minimal perpotongan yang menghasilkan nilai minimal perpotongan sebesar 15% dan jumlah minimal keypoint berkesesuaian antar kedua citra yang dapat digabungkan adalah 5. Variasi rotasi untuk berapapun perbedaan sudut antara kedua citra tetap dapat digabungkan. Variasi skala minimal citra yang dapat digabungkan adalah skala citra yang diperkecil hingga 75% dari ukuran citra aslinya, dan skala citra yang diperbesar hingga 600% dari ukuran aslinya untuk maksimal variasi perbesaran skala. Kata kunci—Foto udara, Penggabungan gambar, SURF, FLANN, RANSAC AbstractPrototype of stitching aerial photograph UAV using SURF algorithm is a system that is designed to stitch the image. The image is generated imagery UAV fixed-wings. The output of this system is a display image with a wider object. This system is designed to be able to merge aerial images by using SURF algorithm, Flann, RANSAC, and warpPerspective. SURF algorithm is used as a keypoint detector from each of the input images. Flann method to perform keypoint matching is found. RANSAC homography matrix used for the search. WarpPerspective method used for merging the two inputs that have a match keypoint. The test consists of several types of variations such as the intersection of the minimal variation, variation in scale and rotation variations. Variation that produces intersection minimum value of 15% and a minimum number of keypoint accords between the two images can be combined is 5. Variation of rotation to any angle difference between the two images can still be combined. Minimum scale variations which can be combined image is the image scale is reduced to 75% of the size of the original image, and the image scale is enlarged to 600% of its original size to a maximum variation of magnification scale. Keywords—Aerial Photograph, Stitching Images, SURF, FLANN, RANSAC

단일 카메라를 이용한 이미지 유사도 비교 기반의 사용자 위치추정

In this paper, user-position estimation method is proposed by using a single camera for both indoor and outdoor environments. Conventionally, the GPS of RF-based estimation methods have been widely studied in the literature for outdoor and indoor environments, respectively. Each method is useful only for indoor or outdoor environment. In this context, this study adopts a vision-based approach which can be commonly applicable to both environments. Since the distance or position cannot be extracted from a single still image, the reference images pro-stored in image database are used to identify the current position from the single still image captured by a single camera. The reference image is tagged with its captured position. To find the reference image which is the most similar to the current image, the SURF algorithm is used for feature extraction. The outliers in extracted features are discarded by using RANSAC algorithm. The performance of the proposed method is evaluated for two buildings and their outsides for both indoor and outdoor environments, respectively.