SURF Feature Descriptor Research Articles

Automatic Apple Detection and Counting with AD-YOLO and MR-SORT.

In the production management of agriculture, accurate fruit counting plays a vital role in the orchard yield estimation and appropriate production decisions. Although recent tracking-by-detection algorithms have emerged as a promising fruit-counting method, they still cannot completely avoid fruit occlusion and light variations in complex orchard environments, and it is difficult to realize automatic and accurate apple counting. In this paper, a video-based multiple-object tracking method, MR-SORT (Multiple Rematching SORT), is proposed based on the improved YOLOv8 and BoT-SORT. First, we propose the AD-YOLO model, which aims to reduce the number of incorrect detections during object tracking. In the YOLOv8s backbone network, an Omni-dimensional Dynamic Convolution (ODConv) module is used to extract local feature information and enhance the model's ability better; a Global Attention Mechanism (GAM) is introduced to improve the detection ability of a foreground object (apple) in the whole image; a Soft Spatial Pyramid Pooling Layer (SSPPL) is designed to reduce the feature information dispersion and increase the sensory field of the network. Then, the improved BoT-SORT algorithm is proposed by fusing the verification mechanism, SURF feature descriptors, and the Vector of Local Aggregate Descriptors (VLAD) algorithm, which can match apples more accurately in adjacent video frames and reduce the probability of ID switching in the tracking process. The results show that the mAP metrics of the proposed AD-YOLO model are 3.1% higher than those of the YOLOv8 model, reaching 96.4%. The improved tracking algorithm has 297 fewer ID switches, which is 35.6% less than the original algorithm. The multiple-object tracking accuracy of the improved algorithm reached 85.6%, and the average counting error was reduced to 0.07. The coefficient of determination R2 between the ground truth and the predicted value reached 0.98. The above metrics show that our method can give more accurate counting results for apples and even other types of fruit.

An efficient technique for object recognition using Shi-Tomasi corner detection algorithm

An efficient feature detection algorithm and image classification is a very crucial task in computer vision system. There are various state-of-the-art feature detectors and descriptors available for an object recognition task. In this paper, the authors have compared the performance of Shi-Tomasi corner detector with SIFT and SURF feature descriptors and evaluate the performance of Shi-Tomasi in combination with SIFT and SURF feature descriptors. To make the computations faster, authors have reduced the size of features computed in all cases by applying locality preserving projection methodology. Features extracted using these algorithms are further classified with various classifiers like K-NN, decision tree and random forest. For experimental work, a public dataset, namely Caltech-101 image dataset, is considered in this paper. This dataset comprises of 101 object classes. These classes have further contained many images. Using a combination of Shi-Tomasi, SIFT and SURF features, the authors have achieved a recognition accuracy of 85.9%, 80.8% and 74.8% with random forest, decision tree and K-NN classifier, respectively. In this paper, the authors have also computed true positive rate, false positive rate and area under curve in all cases. Finally, the authors have applied the adaptive boosting methodology to improve the recognition accuracy. Authors have reported improved recognition accuracy of 86.4% using adaptive boosting with random forest classifier and a combination of Shi-Tomasi, SIFT and SURF features.

A Novel Hybrid Discrete Cosine Transform Speeded Up Robust Feature-Based Secure Medical Image Watermarking Algorithm

With the advancement of networks and multimedia, digital watermarking technology has received worldwide attention as an effective method of copyright protection. Improving the anti-geometric attack ability of digital watermarking algorithms using image feature-based algorithms have received extensive attention. This paper proposes a novel robust watermarking algorithm based on SURF-DCT perceptual hashing (Speeded Up Robust Features and Discrete Cosine Transform), namely blind watermarking. We design and implement a meaningful binary watermark embedding and extraction algorithm based on the SURF feature descriptor and discrete-cosine transform domain digital image watermarking algorithm. The algorithm firstly uses the affine transformation with a feature matrix and chaotic encryption technology to preprocess the watermark image, enhance the confidentiality of the watermark, and perform block and DCT coefficients extraction on the carrier image, and then uses the positive and negative quantization rules to modify the DCT coefficients. The embedding of the watermark is completed, and the blind extraction of the watermark realized. Correlation values are more than 90% in most of the attacks. It provides better results against different noise attacks and also better performance against rotation. Transparency and high computational efficiency, coupled with dual functions of copyright protection and content authentication, is the advantage of the proposed algorithm.

A real-time approach for automatic defect detection from PCBs based on SURF features and morphological operations

This paper presents an automatic inspection approach for Printed Circuit Boards (PCBs) with accurate determination of the fault location and identification of the fault type. This approach depends on several digital image processing techniques including registration, filtering, foreground segmentation, mathematical morphological operations, subtraction, feature extraction, and component matching. The Speeded Up Robust Feature extraction (SURF) technique is used for two purposes: registration of the PCB to be checked to a reference PCB and detection of feature points of each missing component on the PCB that is localized from the subtraction process from the reference PCB. Operation is performed on the hue component of the color PCB images. A dictionary is first built for all possible components on the available PCBs with SURF feature descriptors, and hence if a missing item is detected on a PCB during the inspection process, the SURF feature descriptors for features extracted from the difference between the tested and reference PCBs at the position of the lost component are matched with those in the built dictionary or database. A distance metric is used in the matching process. The importance of the proposed approach lies in its ability to build a dictionary of feature descriptors for all possible components in a diversity of PCBs and its ability to localize and identify the missing components regardless of the PCB position, rotation, or type. All operations are formulated in a Graphical User Interface (GUI) using MATLAB environment.

Real Time Monocular Visual Odometry Using Hybrid Features and Distance Ratio for Scale Estimation

Real time dead reckoning navigation is important for supplying information of the current position of an autonomous mobile robot to complete its task, especially in certain areas such as hazardous and GPS-denied areas. Monocular visual odometry is a good choice as it is one of the dead reckoning navigation method which only uses single camera. For real time task, visual odometry requires fast feature extraction without ignoring its accuracy. Therefore, we propose the usage of a hybrid feature, i.e. Censure feature detector and upright SURF feature descriptor, as feature extraction. Yet, the scale ambiguity for the monocular visual odometry becomes a challenging problem. Without additional information from other sensors, estimating the scale is solely the only way. In our proposed work, distance ratio is employed to tackle such problems. Experimental results show the performance of the designed algorithm. A real example of running the proposed algorithm under an embedded device is also provided for demonstrating its real time capability.

Feature description with SIFT, SURF, BRIEF, BRISK, or FREAK? A general question answered for bone age assessment

Solving problems in medical image processing is either generic (being applicable to many problems) or specific (optimized for a certain task). For example, bone age assessment (BAA) on hand radiographs is a frequent but cumbersome task for radiologists. For this problem, many specific solutions have been proposed. However, general-purpose feature descriptors are used in many computer vision applications. Hence, the aim of this study is (i) to compare the five leading keypoint descriptors on BAA, and, in doing so, (ii) presenting a generic approach for a specific task. Two methods for keypoint selection were applied: sparse and dense feature points. For each type, SIFT, SURF, BRIEF, BRISK, and FREAK feature descriptors were extracted within the epiphyseal regions of interest (eROI). Classification was performed using a support vector machine. Reference data (1101 radiographs) of the University of Southern California was used for 5-fold cross-validation. The data was grouped into 30 classes representing the bone age range of 0–18 years. With a mean error of 0.605 years, dense SIFT gave best results and outperforms all published methods. The accuracy was 98.36% within the range of 2 years. Dense SIFT represents a generic method for a specific question.

Feature Extractionusing ORB-RANSAC for Face Recognition

Abstract Face Recognition is one of the most thriving and cutting- edge fields of research that stands unwaveringly as the most critically challenging problems in the domain of Computer Vision. In the design of an efficient FR system, the potency of the feature descriptor dictates the proficiency of the recognition performance. The effective employment of FR systems in handheld and mobile devices is hindered by their lack of GPU acceleration, which considerably limits their computational capabilities and furthermore, the prevalent SURF and SIFT feature descriptors are computationally expensive and hence are not viable for low-powered devices. To that end, ORB, a cost effective feature descriptor, has been effective for such devices. Hence, in this paper, we propose a novel technique that utilizes ORB, and in turn, address a few of its crucial inadequacies by incorporating RANSAC as a post-processing step to remove redundant key-points and noise (in the form of outliers) and hence improve ORB's efficacy to proffer a robust system for facial image recognition, that has improved accuracy than the prevalent state-of-the-art methods such as SIFT-RANSAC and SURF-RANSAC.

ORB-PCA Based Feature Extraction Technique for Face Recognition

Abstract Face Recognition is one of the most prevalent fields in the domain of Computer Vision and the problems pertaining to it are very challenging. The potency of FR systems in biometric authentication systems is slowed down by their processing speed constraints, which substantially limits their computational capabilities and furthermore, the prevalent SURF and SIFT feature descriptors are computationally complex and demand very high level hardware requirements for processing. Off late, ORB, a cost effective feature descriptor, has been effective for such systems. Hence, in this paper, we propose a novel technique that utilizes ORB, and in turn, incorporate PCA as a post-processing step to reduce the dimensionality of the descriptors and hence improve ORB's authentication performance over the widespread methods such as SIFT-PCA and SURF-PCA.