Image Feature Detection Research Articles

2-Levels of clustering strategy to detect and locate copy-move forgery in digital images

Understanding is considered a key purpose of image forensic science in order to find out if a digital image is authenticated or not. It can be a sensitive task in case images are used as necessary proof as an impact judgment. it’s known that There are several different manipulating attacks but, this copy move is considered as one of the most common and immediate one, in which a region is copied twice in order to give different information about the same scene, which can be considered as an issue of information integrity. The detection of this kind of manipulating has been recently handled using methods based on SIFT. SIFT characteristics are represented in the detection of image features and determining matched points. A clustering is a key step which always following SIFT matching in-order to classify similar matched points to clusters. The ability of the image forensic tool is represented in the assessment of the conversion that is applied between the two duplicated images of one region and located them correctly. Detecting copy-move forgery is not a new approach but using a new clustering approach which has been purposed by using the 2-level clustering strategy based on spatial and transformation domains and any previous information about the investigated image or the number of clusters need to be created is not necessary. Results from different data have been set, proving that the proposed method is able to individuate the altered areas, with high reliability and dealing with multiple cloning.

FEATURE MATCHING OF HISTORICAL IMAGES BASED ON GEOMETRY OF QUADRILATERALS

Abstract. This contribution shows an approach to match historical images from the photo library of the Saxon State and University Library Dresden (SLUB) in the context of a historical three-dimensional city model of Dresden. In comparison to recent images, historical photography provides diverse factors which make an automatical image analysis (feature detection, feature matching and relative orientation of images) difficult. Due to e.g. film grain, dust particles or the digitalization process, historical images are often covered by noise interfering with the image signal needed for a robust feature matching. The presented approach uses quadrilaterals in image space as these are commonly available in man-made structures and façade images (windows, stones, claddings). It is explained how to generally detect quadrilaterals in images. Consequently, the properties of the quadrilaterals as well as the relationship to neighbouring quadrilaterals are used for the description and matching of feature points. The results show that most of the matches are robust and correct but still small in numbers.

Interview

Dr Pouya Bolourchi from Final International University, Turkey, talks to Electronics Letters about his paper ‘Entropy score based feature selection for moment based SAR image classification’, page 593. Pouya Bolourchi Feature detection and classification of images is one the most fascinating topics in image processing, which is why I have been working on various topics in this subject field for more than five years. My recent research focus has included remote sensing, especially in the classification of synthetic aperture radar (SAR) images of ground vehicles. The increasing number of databases, as a result of enhanced SAR acquisition systems, has turned this field into a major research direction to follow. I believe that the industrial demand for remote sensing will exponentially grow and there will be more opportunities to improve the processing and analysis of SAR images. Target recognition consists of three major phases, including feature extraction, feature selection and classification. My research contributes to the first two phases. Instead of proposing a new feature extraction technique, this Letter focuses on the selection of the most discriminative features that are generated by different feature extraction algorithms. The proposed entropy score-based feature selection method is effective for maximising the level of discrimination between different classes by incorporating only the most discriminant features in the selected feature subset. In the first phase, we applied 12 moment methods for feature extraction. The extracted features from different moment methods can be fused to form a high-dimensional feature vector. However, reducing the dimensionality and increasing the performance of such a vector are two difficult challenges to overcome. Therefore, we propose a unique approach, which selects the top moments after sorting the 12 moments using an entropy score-driven feature selection. The approach is powered by an entropy-score metric, which is introduced to select the feature vectors that maximise the entropy ratio of total sample population and the average of within-class populations. Having a high entropy score guarantees a feature vector with high-scattered total population and low class-specific deviation, which can be used as a strong class-separation score. The main task in automatic target recognition is to enhance recognition. SAR with high-resolution images is superior to optical images, due to their robust ability to work in all weather conditions, during day and night. However, extracting many features from such a high-resolution image is a challenging task. Furthermore, positional variations of the target, due to angle variations in airborne radar signals, generate images that are not scaled. For recognition enhancement, the proposed entropy-score method follows the moment-based feature extraction process to perform feature selection. The biggest challenge was to perform dimensionality reduction, while increasing the discrimination between classes. In the short term, the proposed algorithm can be effectively applied to many different fields, ranging from hyperspectral images to biomedical images, in order to perform efficient feature selection after forming a high-dimensional feature space. In the longer term, the proposed entropy-score-based method can be developed to increase the discrimination between classes considered. The mathematical foundations of the proposed method will also be explored to provide theoretical alternatives in practical applications to provide possible improvements. Currently, I am working on Alzheimer's disease detection in 3D Magnetic Resonance Imaging (MRI) images. Moment methods in the feature extraction stage and entropy-score in feature selection phase can be adopted in this field. In the case of MRI images, dimensionality reduction must be performed for effective classification. However, I believe that detection of Alzheimer's disease without any treatment is meaningless, so instead of bi-level classification, we plan to determine the level of Alzheimer's disease for possible early treatment. Detection and classification research is growing rapidly. The application of classification varies from face recognition to security, to medical image diagnosis. There is a big hope that image processing will contribute to medical science for the diagnosis of difficult diseases, like cancer and Alzheimer's disease, over the coming years. The developing role of image processing in medical science is obvious in recent years and is expected to grow rapidly in the coming decades.

BAFS: Bundle Adjustment With Feature Scale Constraints for Enhanced Estimation Accuracy

We propose to incorporate within bundle adjustment (BA) a new type of constraint that uses feature scale information, leveraging the scale invariance property of typical image feature detectors (e.g., SIFT). While feature scales play an important role in image matching, they have not been utilized thus far for estimation purposes in a BA framework. Our approach exploits the already-available feature scale information and uses it to enhance the accuracy of BA, especially along the optical axis of the camera in a monocular setup. Importantly, the mentioned feature scale constraints can be formulated on a frame to frame basis and do not require loop closures. We study our approach in synthetic environments and the real-imagery KITTI dataset, demonstrating significant improvement in positioning error.

In Situ Visualization Measurement of Flat Plate Ablation in High-Temperature Gas Flow

In this work, we develop an optoelectronic system for in situ observation and measurement in hypervelocity flows. The system has the advantages of strong radiation resistance and self-adaptive exposure time of the cameras. Thermal ablation test using flat plate thermal protection system material was carried out in an arc jet. Real-time ablation images were captured and analyzed to understand the thermal ablation mechanism. Through the modified algorithms of particle image velocity (PIV) and image feature detection, the surface recession rate and the velocity distribution of the melted droplets flowing on the sample surface were obtained. The experimental results demonstrate vast potential for using this in situ measuring technique in various engineering applications. Finally, the formation and merging of the melted droplets was analyzed based on energy theory, and the numerical simulation results showed good agreement with the actual experimental results.

Feature Detection of Curve Traffic Sign Image on The Bandung - Jakarta Highway

Unsealed roadside and problems with the road surface are common causes of road crashes, particularly when those are combined with curves. Curve traffic sign is an important component for giving early warning to driver on traffic, especially on high-speed traffic like on the highway. Traffic sign detection has became a very interesting research now, and in this paper will be discussed about the detection of curve traffic sign. There are two types of curve signs are discussed, namely the curve turn to the left and the curve turn to the right and the all data sample used are the curves taken / recorded from some signs on the Bandung - Jakarta Highway. Feature detection of the curve signs use Speed Up Robust Feature (SURF) method, where the detected scene image is 800x450. From 45 curve turn to the right images, the system can detect the feature well to 35 images, where the success rate is 77,78%, while from the 45 curve turn to the left images, the system can detect the feature well to 34 images and the success rate is 75,56%, so the average accuracy in the detection process is 76,67%. While the average time for the detection process is 0.411 seconds.

Privacy-Preserving Image Processing in the Cloud

Millions of private images are generated in various digital devices every day. The consequent massive computational workload makes people turn to cloud computing platforms for their economical computation resources. Meanwhile, the privacy concerns over the sensitive information contained in outsourced image data arise in public. In fact, once uploaded to cloud, the security and privacy of the image content can only presume upon the reliability of the cloud service providers. Lack of assuring security and privacy guarantees becomes the main barrier to further deployment of cloud-based image processing systems. This paper studies the design targets and technical challenges lie in constructing cloud-based privacy-preserving image processing system. We explore various image processing tasks, including image feature detection, digital watermarking, content-based image search. The state-of-the-art techniques, including secure multiparty computation, and homomorphic encryption are investigated. A detailed taxonomy of the problem statement and the corresponding solutions is provided

Combining Deep and Handcrafted Image Features for Presentation Attack Detection in Face Recognition Systems Using Visible-Light Camera Sensors.

Although face recognition systems have wide application, they are vulnerable to presentation attack samples (fake samples). Therefore, a presentation attack detection (PAD) method is required to enhance the security level of face recognition systems. Most of the previously proposed PAD methods for face recognition systems have focused on using handcrafted image features, which are designed by expert knowledge of designers, such as Gabor filter, local binary pattern (LBP), local ternary pattern (LTP), and histogram of oriented gradients (HOG). As a result, the extracted features reflect limited aspects of the problem, yielding a detection accuracy that is low and varies with the characteristics of presentation attack face images. The deep learning method has been developed in the computer vision research community, which is proven to be suitable for automatically training a feature extractor that can be used to enhance the ability of handcrafted features. To overcome the limitations of previously proposed PAD methods, we propose a new PAD method that uses a combination of deep and handcrafted features extracted from the images by visible-light camera sensor. Our proposed method uses the convolutional neural network (CNN) method to extract deep image features and the multi-level local binary pattern (MLBP) method to extract skin detail features from face images to discriminate the real and presentation attack face images. By combining the two types of image features, we form a new type of image features, called hybrid features, which has stronger discrimination ability than single image features. Finally, we use the support vector machine (SVM) method to classify the image features into real or presentation attack class. Our experimental results indicate that our proposed method outperforms previous PAD methods by yielding the smallest error rates on the same image databases.

FEATURE EXTRACTION OF COLON AND RECTUM CANCER FROM MRI IMAGES

Feature extraction, the third major component plays a vital role on the final expected outcome of the application using digital image processing. Feature plays a very significant role in the area of digital image processing. The image feature detection and extraction are based on the colour, shape and texture feature categories. Colour feature refers to the presence of one of the colours either in RGB space (Red, Green, Blue colour model) or in HSxspace (Hue, Saturation, Value/Brightness model) in the particular area of interest. Especially in medical imaging, the internal parts will have a specific colour slightly varying in hue. When the scan modality provides a colour image, colour feature can be used for processing. The structure of the organs and its parts if visible clearly, then the shape parameter can be used for classification of images. The knowledge of the domain experts will enable to find the structure as what and where it is located. Similarity in shape will help in identifying the objects. Irrespective of colour and shape, texture parameter plays an important role in classification problems. In the absence of colour and shape, the only available feature is the textural features of the object. It looks into the properties related with intensity values of each pixel. This paper proposes the textural feature extraction techniques that best suits for retrieving the properties of colon and rectum cancer MRI images which can be further processed for classification.

Energy Spectrum CT Image Detection Based Dimensionality Reduction with Phase Congruency.

The image feature detection is widely used in image registration, image stitching and object recognition. The feature detection algorithm can be applied to the detection of artificial images, and can be used to detect the energy spectrum CT image. A new algorithm of phase consistency detection based on dimensionality reduction is proposed in this paper. We mainly focus on the phase congruency of the spectral CT images in the paper and try to use dimensionality reduction to integrate the information of phase congruency detected in the image. The experimental results show that the algorithm can detect the energy spectrum CT image with clear edge and contour, which is beneficial to the subsequent processing. Meanwhile, the algorithm presented is more effective in diagnosis of disease for medical professionals.

Performance Characterization of Image Feature Detectors in Relation to the Scene Content Utilizing a Large Image Database

Selecting the most suitable local invariant feature detector for a particular application has rendered the task of evaluating feature detectors a critical issue in vision research. Although the literature, offers a variety of comparison works focusing on performance evaluation of image feature detectors under several types of image transformations, the influence of the scene content on the performance of local feature detectors has received little attention so far. This paper, aims to bridge this gap with a new framework for determining the type of scenes which maximize and minimize the performance of detectors in terms of repeatability rate. The results are presented for several state-of-the-art feature detectors that have been obtained using a large image database of 20482 images under JPEG compression, uniform light and blur changes with 539 different scenes captured from real-world scenarios. These results provide new insights into the behavior of feature detectors.

A Survey of 3D Image Navigation and High Precision Dynamic Registration in Minimally Invasive Surgery

Image registration is one of the fundamental research areas in image processing and computer vision, which plays a crucial and irreplaceable role in the field of image navigation. In this paper, an overview of 3D image navigation and high precision dynamic registration in minimally invasive surgery are presented. Firstly, the main principles of modern medical image registration techniques were introduced, and its new characteristics in modern image survey navigation were analyzed. Secondly, the feature-based and gray-based image matching methods are studied, which are the basic idea behind the image registration technique. Meanwhile, the advantages and disadvantages of these two image matching algorithms itself and the shortcomings in the application of image registration were also analyzed. Finally, the superiority of noise robust and high-resolution image feature detection methods in the field of image registration have been analyzed, which may be a valuable direction of future research.

Image feature detection using an improved implementation of maximally stable extremal regions for augmented reality applications

ABSTRACTAugmented Reality for all practical purposes requires extensive computation, accurate view alignment and real-time performance. To address some of these limitations, an improved method of feature detection is proposed using Maximally Stable Extremal Regions. The approach used for feature detection extracts the regions of interest using a true flood fill approach for building and maintaining the component tree. This approach has true worst-case linear complexity (Linear-MSER). In the present work, Linear-MSER is implemented at multiple scales of an image in order to increase the affine invariance properties of the detector (MSLinear-MSER).The two detectors, Linear-MSER and MSLinear-MSER, are then combined separately with Scale Invariant Feature Transform and Speeded-Up Robust Feature descriptors for performance comparison. Performance evaluation is done under varying imaging conditions like changes in viewpoint, scale, blur, illumination and JPEG compression. Results show that, MSLinear-MSER+SIFT performs best in terms of time-complexity and number of keypoint matches when executed at six octaves and five levels. This observation is true for all image-sets taken into consideration, containing images that are affine transformed in one way or other. To exhibit the efficiency of MSLinear-MSER+SIFT, a prototype of an AR system is also developed and discussed in this article using this approach.

Accurate Diffraction Peak Identification for Scanning Electron Nanodiffraction Based on Automated Image Processing and Feature Detection

Accurate Diffraction Peak Identification for Scanning Electron Nanodiffraction Based on Automated Image Processing and Feature Detection

THE EFFECT OF IMAGE ENHANCEMENT METHODS DURING FEATURE DETECTION AND MATCHING OF THERMAL IMAGES

Abstract. A successful image matching is essential to provide an automatic photogrammetric process accurately. Feature detection, extraction and matching algorithms have performed on the high resolution images perfectly. However, images of cameras, which are equipped with low-resolution thermal sensors are problematic with the current algorithms. In this paper, some digital image processing techniques were applied to the low-resolution images taken with Optris PI 450 382 x 288 pixel optical resolution lightweight thermal camera to increase extraction and matching performance. Image enhancement methods that adjust low quality digital thermal images, were used to produce more suitable images for detection and extraction. Three main digital image process techniques: histogram equalization, high pass and low pass filters were considered to increase the signal-to-noise ratio, sharpen image, remove noise, respectively. Later on, the pre-processed images were evaluated using current image detection and feature extraction methods Maximally Stable Extremal Regions (MSER) and Speeded Up Robust Features (SURF) algorithms. Obtained results showed that some enhancement methods increased number of extracted features and decreased blunder errors during image matching. Consequently, the effects of different pre-process techniques were compared in the paper.

A variant of the Hough Transform for the combined detection of corners, segments, and polylines

The Hough Transform (HT) is an effective and popular technique for detecting image features such as lines and curves. From its standard form, numerous variants have emerged with the objective, in many cases, of extending the kind of image features that could be detected. Particularly, corner and line segment detection using HT has been separately addressed by several approaches. To deal with the combined detection of both image features (corners and segments), this paper presents a new variant of the Hough Transform. The proposed method provides an accurate detection of segment endpoints, even if they do not correspond to intersection points between line segments. Segments are detected from their endpoints, producing not only a set of isolated segments but also a collection of polylines. This provides a direct representation of the polygonal contours of the image despite imperfections in the input data such as missing or noisy feature points. It is also shown how this proposal can be extended to detect predefined polygonal shapes. The paper describes in detail every stage of the proposed method and includes experimental results obtained from real images showing the benefits of the proposal in comparison with other approaches.

Affine scale space: an affine invariant image structure to promote the detection of correspondences from stereo images

Calculating the geometry relationship by the positions of the correspondences from stereo images is a fundamental method to obtain the depth information. Such a method was quite widespread and popular thanks to its efficiency and easily accessed implementation. General speaking, the more density of the correspondences are, the more precisely the depth information can be calculated. Theoretically, a sufficient strengthened correspondences match algorithm can be utilized for depth information calculation under any circumstances. Unfortunately, the updated image feature detectors are all sensitive to the view point changes: with the rising of the stereo images' view angle, the number of matched features drastically reduced, resulting in the number of matched features not adequate to cover every details of the stereo images. This disadvantages of feature detections in practice hampers its application for the depth calibration. To tackle the sensitive of view point to stereo images, in this paper, we will propose an affine invariant affine scale space structure, which is more robust to detect the correspondences from stereo images. The purpose of affine scale space is to create a more general approach to the affine invariant image scale representation by modifying the corresponding Gaussian filters in order to cope with the specific change of view point. The affine adaptation of the scale space is to retain a linear relationship with the transiting of the view point. With linear relationship, the affine scale space can be established as a more general approach for the detection of correspondences from stereo images. With a better correspondences detection, a more precise depth information can be made.

Comparison of Local Feature Extraction Paradigms Applied to Visual SLAM

The detection and description of locally salient regions is one of the most widely used low-level processes in modern computer vision systems. The general approach relies on the detection of stable and invariant image features that can be uniquely characterized using compact descriptors. Many detection and description algorithms have been proposed, most of them derived using different assumptions or problem models. This work presents a comparison of different approaches towards the feature extraction problem, namely: (1) standard computer vision techniques; (2) automatic synthesis techniques based on genetic programming (GP); and (3) a new local descriptor based on composite correlation filtering, proposed for the first time in this paper. The considered methods are evaluated on a difficult real-world problem, vision-based simultaneous localization and mapping (SLAM). Using three experimental scenarios, results indicate that the GP-based methods and the correlation filtering techniques outperform widely used computer vision algorithms such as the Harris and Shi-Tomasi detectors and the Speeded Up Robust Features descriptor.

FPGA accelerator for real-time SIFT matching with RANSAC support

Scale-Invariant Feature Transform (SIFT) has been considered as one of the more robust techniques for the detection and matching of image features. However, SIFT is computationally demanding and it is rarely used when real time operation is required. In this paper, a complete FPGA architecture for feature matching in consecutive video frames is proposed. Procedures of SIFT detection and description are fully parallelized. At every clock cycle, the current pixel in the pipeline is tested and if it is a SIFT feature, its descriptor is extracted. Furthermore, every detected feature in the current frame is matched with one among the stored features of the previous frame, using a moving window, without violating pixel pipelining. False matches are rejected using random sample consensus (RANSAC) algorithm. Each RANSAC run lasts for as many clock cycles as the number of the selected random samples. The architecture was verified in the DE2i-150 development board. In the target hardware, maximum supported clock frequency is 25MHz and the architecture is capable to process more than 81fps, considering image resolution of 640 × 480pixels.

A Generic Framework for Assessing the Performance Bounds of Image Feature Detectors

Since local feature detection has been one of the most active research areas in computer vision during the last decade and has found wide range of applications (such as matching and registration of remotely sensed image data), a large number of detectors have been proposed. The interest in feature-based applications continues to grow and has thus rendered the task of characterizing the performance of various feature detection methods an important issue in vision research. Inspired by the good practices of electronic system design, a generic framework based on the repeatability measure is presented in this paper that allows assessment of the upper and lower bounds of detector performance and finds statistically significant performance differences between detectors as a function of image transformation amount by introducing a new variant of McNemar’s test in an effort to design more reliable and effective vision systems. The proposed framework is then employed to establish operating and guarantee regions for several state-of-the art detectors and to identify their statistical performance differences for three specific image transformations: JPEG compression, uniform light changes and blurring. The results are obtained using a newly acquired, large image database (20,482 images) with 539 different scenes. These results provide new insights into the behavior of detectors and are also useful from the vision systems design perspective. Finally, results for some local feature detectors are presented for a set of remote sensing images to showcase the potential and utility of this framework for remote sensing applications in general.