Hu Moments Research Articles

Plant Recognition Based on Multi-Feature and Locality Preserving Projections Fusion Algorithm

For the demand of leaf recognition under rotation condition. This paper propose a method to extract the multi-feature extraction and Locally Preserving Projection (LPP) fusion algorithm. First, the texture feature of the leaf is extracted by Local Binary Pattern (LBP) algorithm based on leaf block. Then, in order to improve the speed of feature classification recognition, this paper adopts LPP algorithm for feature dimension reduction of the high dimensional of LBP. In addition, combining leaf texture feature LBP with shape feature (Hu Moment invariants) after dimension reduction, shaping into the comprehensive characteristics of plant leaves. Finally, Support Vector Machine (SVM) classifier is used to establish classifier for classification and identification of plant leaves, the results of experiment have verified the validity of the algorithm in this paper.

Beyond Features for Recognition: Human-Readable Measures to Understand Users’ Whole-Body Gesture Performance

ABSTRACTUnderstanding users’ whole-body gesture performance quantitatively requires numerical gesture descriptors or features. However, the vast majority of gesture features that have been proposed in the literature were specifically designed for machines to recognize gestures accurately, which makes those features exclusively machine-readable. The complexity of such features makes it difficult for user interface designers, non-experts in machine learning, to understand and use them effectively (see, for instance, the Hu moment statistics or the Histogram of Gradients features), which reduces considerably designers’ available options to describe users’ whole-body gesture performance with legible and easily interpretable numerical measures. To address this problem, we introduce in this work a set of 17 measures that user interface practitioners can readily employ to characterize users’ whole-body gesture performance with human-readable concepts, such as area, volume, or quantity. Our measures describe (1) spatial characteristics of body movement, (2) kinematic performance, and (3) body posture appearance for whole-body gestures. We evaluate our measures on a public dataset composed of 5,654 gestures collected from 30 participants, for which we report several gesture findings, e.g., participants performed body gestures in an average volume of space of 1.0 m3, with an average amount of hands movement of 14.6 m, and a maximum body posture diffusion of 5.8 m. We show the relationship between our gesture measures and recognition rates delivered by a template-based Nearest-Neighbor whole-body gesture classifier implementing the Dynamic Time Warping dissimilarity function. We also release BOGArT, the Body Gesture Analysis Toolkit, that automatically computes our measures. This work will empower researchers and practitioners with new numerical tools to reach a better understanding of how users perform whole-body gestures and thus, to use this knowledge to inform improved designs of whole-body gesture user interfaces.

Automatic target recognition system for unmanned aerial vehicle via backpropagation artificial neural network

PurposeThe purpose of this paper is to propose a novel target automatic recognition method for unmanned aerial vehicle (UAV), which is based on backpropagation – artificial neural network (BP-ANN) algorithm, with the objective of optimizing the structure of backpropagation network, to increase the efficiency and decrease the recognition time. A hardware-in-the-loop system for UAV target automatic recognition is also developed.Design/methodology/approachThe hybrid model of BP-ANN structure is established for aircraft automatic target recognition. This proposed method identifies controller parameters and reduces the computational complexity. Approaching speed of the network is faster and recognition accuracy is higher. This kind of network combines or better fuses the advantages of backpropagation artificial neural algorithm and Hu moment. with advantages of two networks and improves the speed and accuracy of identification. Finally, a hardware-in-the-loop system for UAV target automatic recognition is also developed.FindingsThe double hidden level backpropagation artificial neural can easily increase the speed of recognition process and get a good performance for recognition accuracy.Research limitations/implicationsThe proposed backpropagation artificial neural algorithm can be ANN easily applied to practice and can help the design of the aircraft automatic target recognition system. The standard backpropagation algorithm has some obvious drawbacks, namely, converging slowly and falling into the local minimum point easily. In this paper, an improved algorithm based on the standard backpropagation algorithm is constructed to make the aircraft target recognition more practicable.Originality/valueA double hidden levels backpropagation artificial neural algorithm is presented for automatic target recognition system of UAV.

Classification of Alzheimer Disease based on Normalized Hu Moment Invariants and Multiclassifier

There is a great benefit of Alzheimer disease (AD) classification for health care application. AD is the most common form of dementia. This paper presents a new methodology of invariant interest point descriptor for Alzheimer disease classification. The descriptor depends on the normalized Hu Moment Invariants (NHMI). The proposed approach deals with raw Magnetic Resonance Imaging (MRI) of Alzheimer disease. Seven Hu moments are computed for extracting images’ features. These moments are then normalized giving new more powerful features that highly improve the classification system performance. The moments are invariant which is the robustness point of Hu moments algorithm to extract features. The classification process is implemented using two different classifiers, K-Nearest Neighbors algorithm (KNN) and Linear Support Vector Machines (SVM). A comparison among their performances is investigated. The results are evaluated on Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. The best classification accuracy is 91.4% for KNN classifier and 100% for SVM classifier.

3D Human Action Recognition using Hu Moment Invariants and Euclidean Distance Classifier

This paper presents a new model of scale, rotation, and translations invariant interest point descriptor for human actions recognition. The descriptor, HMIV (Hu Moment Invariants on Videos) is used for solving surveillance camera recording problems under different conditions of side, position, direction and illumination. The proposed approach deals with raw input human action video sequences. Seven Hu moments are computed for extracting human action features and for storing them in a 1D vector which is constringed as one mean value for all the frames’ moments. The moments are invariant to scale, translation, or rotation, which is the robustness point of Hu moments algorithm. The experiments are evaluated using two different datasets; KTH and UCF101. The classification process is executed by calculating the Euclidean distance between the training and testing datasets. Human action with minimum distance will be selected as the winner matching action. The maximum classification accuracy in this work is 93.4% for KTH dataset and 92.11% for UCF101.

Indian Classical Dance Classification with Adaboost Multiclass Classifier on Multifeature Fusion

Extracting and recognizing complex human movements from unconstraint online video sequence is an interesting task. In this paper the complicated problem from the class is approached using unconstraint video sequences belonging to Indian classical dance forms. A new segmentation model is developed using discrete wavelet transform and local binary pattern (LBP) features for segmentation. A 2D point cloud is created from the local human shape changes in subsequent video frames. The classifier is fed with 5 types of features calculated from Zernike moments, Hu moments, shape signature, LBP features, and Haar features. We also explore multiple feature fusion models with early fusion during segmentation stage and late fusion after segmentation for improving the classification process. The extracted features input the Adaboost multiclass classifier with labels from the corresponding song (tala). We test the classifier on online dance videos and on an Indian classical dance dataset prepared in our lab. The algorithms were tested for accuracy and correctness in identifying the dance postures.

SAR target recognition based on active contour without edges

A new method for synthetic aperture radar (SAR) target recognition is proposed. This method is accomplished via the combination of active contour without edges, Hu invariant moments and support vector machine (SVM) classifier. Image segmentation is performed by using active contour without edges. Then seven Hu moments are extracted and normalized as feature vectors. Finally, the SVM classifier is employed for data training and testing by means of MSTAR SAR images. To verify the performance of the proposed method, the traditional active contour (snakes) is used for comparison. The simulation results confirm the feasibility and accuracy of the proposed method in SAR target recognition.

Full-automatic computer aided system for stem cell clustering using content-based microscopic image analysis

Stem cells are very original cells that can differentiate into other cells, tissues and organs, which play a very important role in biomedical treatments. Because of the importance of stem cells, in this paper we propose a full-automatic computer aided clustering system to assist scientists to explore potential co-occurrence relations between the cell differentiation and their morphological information in phenotype. In this proposed system, a multi-stage Content-based Microscopic Image Analysis (CBMIA) framework is applied, including image segmentation, feature extraction, feature selection, feature fusion and clustering techniques. First, an Improved Supervised Normalized Cuts (ISNC) segmentation algorithm is newly introduced to partition multiple stem cells into individual regions in an original microscopic image, which is the most important contribution in this paper. Then, based on the segmented stem cells, 11 different feature extraction approaches are applied to represent the morphological characteristics of them. Thirdly, by analysing the robustness and stability of the extracted features, Hu and Zernike moments are selected. Fourthly, these two selected features are combined by an early fusion approach to further enhance the properties of the feature representation of stem cells. Finally, k-means clustering algorithm is chosen to classify stem cells into different categories using the fused feature. Furthermore, in order to prove the effectiveness and usefulness of this proposed system, we carry out a series of experiments to evaluate our methods. Especially, our ISNC segmentation obtains 92.4% similarity, 96.0% specificity and 107.8% ration of accuracy, showing the potential of our work.

QR Code Patterns Localization based on Hu Invariant Moments

The widespread utilization of QR code and its coincidence with the swift growth of e-commerce transactions have imposed the computer vision researchers to continuously devise a variety of QR code recognition algorithms. The latter performances are generally limited due to two main factors. Firstly, most of them are computationally expensive because of the implemented feature descriptor complexities. Secondly, the evoked algorithms are often sensitive to pattern geometric deformations. In this paper a robust approach is proposed, in which the architecture is based on three distinct treatments among others: 1) An image quality assessment stage which evaluates the quality of the captured image in consideration that the presence of blur decreases significantly the recognition accuracy. 2) This stage is followed by an image segmentation based on an achromatic filter through which only the regions of interest are highlighted and consequently the execution time is reduced. 3) Finally, the Hu invariant moments technique is used as feature descriptor permitting removing false positives. This technique is implemented to filter out the set of extracted candidate QR code patterns, which have been roughly extracted by a scanning process. The Hu moments descriptor is able to recognize patterns independently of the geometric transformations they undergo. The experiments show that the incorporation of the aforementioned three stages enhances significantly the recognition accuracy along with a notable diminution of processing time. This makes the proposed approach adapted to embedded systems and devices with limited performances.

Robust Gesture Recognition with Kinect Data Acquisition

To realize the gesture recognition of high precision ratio, the gesture recognition method of multi-model data fusion based on Kinect depth image is proposed, to implement the automatic splicing of models. First of all, the feature package model uses the speeded up robust feature (SURF) algorithm to replace the scale invariant feature transform (SIFT) algorithm to extract features, improve the real-time performance. Secondly, Hu moment is introduced to describe the global gesture features, further improving the recognition rate, the ray casting is used finally, and the obtained coordinate information is used to solve the rigid transformation between two point cloud models. Finally, the proposed data fusion method is verified through two experiments, the algorithm in this paper is better than the traditional support vector machine (SVM) method both in real time performance and recognition rate, and obtains better model splicing effect.

A Comparative Study on Efficiency of Classification Techniques with Zone Level Gabor Features towards Handwritten Telugu Character Recognition

Achieving high accuracies in recognition of handwritten text is a challenging research problem and never exhausting. The factors that instill challenges in handwritten character recognition include high degree of variability in writing, script type and the type of documents etc. In this paper, we focus on recognition of handwritten Telugu text commonly found in document images. The character set include all the vowels, consonants and single level vowel consonant clusters chosen in accordance with the commonly used terminology employed in composition pre-printed documents such as admission forms. In this paper, an algorithm is devised that performs zone-based feature extraction of the segmented character images. In the proposed work, the Gabor features are extracted from the character image at zone level and its efficiency is evaluated individually on two different zone representations and entire image at various scales and orientations. The classification and recognition performance is analyzed using nearest neighbor classifier, Naive Bayesian, multi-class SVM and probabilistic neural networks classifiers. The efficiency of the classifiers are also tested with statistical, Histogram of Gradients and Hu moments‟ feature extraction methods and the best accuracy of the system is found to be 84.8% for Gabor features with zone representation 2 and with multi-class SVM classifier.

Improved deep belief networks and multi-feature fusion for leaf identification

Plant identification based on digital leaf images is a hot topic in the automatic classification of plants. However, due to the increase in the number of plant species, the leaf recognition rate is low because the traditional classification methods extract the few characteristics or use the classifiers with simple structures. This paper applied a combination of texture features and shape features for identification. Texture features include local binary patterns, Gabor filters and gray level co-occurrence matrices, while the shape feature vector is modeled using Hu moment invariants and Fourier descriptors. Improved deep belief networks (DBNs) with dropout, which use proportion integration differentiation control (PID) to decrease the reconstruction error in the process of pre-training, are used as the classifiers. The proposed algorithm was tested on the ICL dataset, and the average recognition rate is 93.9% for 220 types of leaves. The experimental results show that the proposed method has a higher recognition rate and is more robust than the traditional methods, and the training process is completed in a shorter time.

Effect of importance sampling on robust segmentation of audio-cough events in noisy environments.

This paper proposes a new cough detection system based on audio signals acquired from conventional smartphones. The system relies on local Hu moments to characterize cough events and a Λ-NN classifier to distinguish cough events from non-cough ones (speech, laugh, sneeze, etc.) and noisy sounds. To deal with the unbalance between classes, we employ Distinct-Borderline2 Synthetic Minority Oversampling Technique and a bespoke cost matrix. The system additionally features a post-processing module to avoid isolated false negatives and, this way, increases sensitivity. Evaluation has been carried out using a database comprising a variety of cough and non-cough events and different types of background noise. In this study, we specifically focused on noise likely to appear when the user is carrying the smartphone in daily activities. Different Signal to Noise Ratio values were tested ranging between -15 and 0 dB. Our experiments confirm that local Hu moments are suitable not only for characterizing cough events but also for coping with noisy environments. Results show a sensitivity of 94.17% and a specificity of 92.16% at -15 dB. Thus, our system shows potential as a reliable and place-ubiquitous monitoring device that helps patients self-manage their own respiratory diseases and avoids unreported or fabricated symptoms.

Morphological analysis of dendrites and spines by hybridization of ridge detection with twin support vector machine.

Dendritic spines are described as neuronal protrusions. The morphology of dendritic spines and dendrites has a strong relationship to its function, as well as playing an important role in understanding brain function. Quantitative analysis of dendrites and dendritic spines is essential to an understanding of the formation and function of the nervous system. However, highly efficient tools for the quantitative analysis of dendrites and dendritic spines are currently undeveloped. In this paper we propose a novel three-step cascaded algorithm–RTSVM— which is composed of ridge detection as the curvature structure identifier for backbone extraction, boundary location based on differences in density, the Hu moment as features and Twin Support Vector Machine (TSVM) classifiers for spine classification. Our data demonstrates that this newly developed algorithm has performed better than other available techniques used to detect accuracy and false alarm rates. This algorithm will be used effectively in neuroscience research.

A Novel Approach in Vehicle Object Classification System with Hybrid of Central and Hu Moment Features using Back Propagation Algorithm

Objectives: To increase the classification accuracy and the performance of the artificial neural network classifier based on hybrid feature extraction. Two different literatures are hybrid together to obtain the better result. Methods/Analysis: Identification and classification of vehicle object system are based on a hybrid feature extraction method and neural classifier. Every image is divided in two equal size 10x10 sub-block. From each sub-block of the image, central moment and geometrical moment features are extracted without pre-processing. The extracted feature vector is normalized and combined together. Normalization is done by using ZScore normalization technique. Then the normalized feature vectors are fed to the Artificial Neural Network (ANN) classifier by using Feed Forward Back Propagation Algorithm (FFBPA) for classifying the vehicle object. Findings: Illinois at Urbana-Champaign (UICI) standard database is used for vehicle object classification. UIUC Dataset contains 500 car images and 500 non-car images with mixed background. The normalized input feature vectors which have been selected are improving the classification accuracy compared with the previous work. It increases the true categorization ratio and decreases the false categorization ratio. The quantity improved performance result shows 95.3% compared with a similar work of various literature methods. Applications/Improvements: This novel method plays a vital role in applications such as vehicle security system, traffic monitoring system etc.

GLOBAL AND LOCAL DESCRIPTOR FOR CBIR AND IMAGE ENHANCEMENT USING MULTI-FEATURE FUSION METHOD

The research is ongoing in CBIR it is getting much popular. In this retrieval of image is done using a technique that searches the necessary features of image. The main work of CBIR is to get retrieve efficient, perfect and fast results.In this algorithm, fused multi-feature for color, texture and figure features. A global and local descriptor (GLD) is proposed in this paper, called Global Correlation Descriptor (GCD) and Discrete Wavelet Transform (DWT), to excerpt color and surface feature respectively so that these features have the same effect in CBIR. In addition, Global Correlation Vector (GCV) and Directional Global Correlation Vector (DGCV) is proposed in this paper which can integrate the advantages of histogram statistics and Color Structure Descriptor (CSD) to characterize color and consistency features respectively. Also, this paper is implemented by Hu moment (HM) for shape feature, it extract 8 moments for image. For the classification process, apply kernel Support vector machine (SVM). The experimental result has computed precision, recall, f_measure and execution time. Also, worked on two datasets: Corel-1000 and Soccer-280.

Multiple Objects Tracking in Surveillance Video Using Color and Hu Moments

Multiple objects tracking finds its applications in many high level vision analysis like object behaviour interpretation and gait recognition. In this paper, a feature based method to track the multiple moving objects in surveillance video sequence is proposed. Object tracking is done by extracting the color and Hu moments features from the motion segmented object blob and establishing the association of objects in the successive frames of the video sequence based on Chi-Square dissimilarity measure and nearest neighbor classifier. The benchmark IEEE PETS and IEEE Change Detection datasets has been used to show the robustness of the proposed method. The proposed method is assessed quantitatively using the precision and recall accuracy metrics. Further, comparative evaluation with related works has been carried out to exhibit the efficacy of the proposed method.

Invariant moments and transform-based unbiased nonlocal means for denoising of MR images

Nonlocal means (NLM) has been applied successfully to medical images degraded with signal independent additive Gaussian noise. However, magnetic resonance images (MRI) are generally corrupted with signal dependent Rician noise which follows Gaussian distribution at high signal-to-noise ratio (SNR) and Rayleigh distribution at low SNR. Considering these properties of the Rician noise, NLM is applied on the squared magnitude images after subtracting the bias term induced during the formation of MR data. In this paper, we propose the use of Zernike moments (ZMs)-based unbiased NLM approach for MRI denoising. ZMs belong to a family of orthogonal rotation invariant moments and possess many useful characteristics such as rotation invariance, minimum information redundancy and better noise resilience. The ability of ZMs to represent edges and fine structures in any orientation and provide better similarity measures make them more suitable for NLM based MRI denoising while preserving the significant details during the restoration process. In addition, we have also investigated the use of non-orthogonal rotation invariant Hu moments and angular radial transform (ART) for MRI denoising. Detailed experiments are conducted using standard clinical data sets to compare the performance of the proposed approaches with the existing approaches.

Determination of the Appropriate Geometric Moment Invariant Functions for Object Recognition

Objectives: Moment invariants have been largely deployed to solve object recognition problems often without analyzing their appropriateness. This paper presents a heuristic method to determine the most appropriate geometric moment invariant functions for object recognition. Methods: The mathematical structure of the moment invariant functions was studied. The elements in the functions that contribute to amplification of noise in the images were identified through derivations. The moment invariant functions were then classified into noise susceptible and noise resilient categories. Sample functions from each of these categories were applied over standard images subjected to different types of noises to appraise the performance of the moment invariant functions. Findings: Research work accomplished using geometric moment invariant functions always showed evidence of false positives and false negatives due to the inherent weakness of some of the moment invariant functions. Most or all of the works using geometric moment invariant functions used only the first seven Hu moment invariants. Upon expansion of the moment invariant functions, it became essential to effectively choose the functions. Existing research explains that the higher the order of the moment functions the more susceptible are they to noise in the images. The research accomplished in this paper proves that not all higher order functions are susceptible to noise. This paper proves that there are higher order geometric functions with smaller exponents which are highly resilient to noise. Applications/Improvements: The research contributed towards identifying the most suitable moment invariant functions for object recognition. This approach helps to minimize false positives and false negatives and choose noise resilient moment invariant functions.

A Vision-Based Precipitation Sensor for Detection and Classification of Hydrometeors

Measuring precipitation is an important part of ground observations of the Earth's atmosphere. Existing systems for this task focus mainly on the hydrometeors' micro-structure (e.g., shape, size, and velocity), but seldom consider to classify them. This paper proposes a new vision-based system for precipitation observation and type recognition (POTR) comprising a single camera and other commercially available components. The system is efficient in terms of energy use and memory requirements by being able to switch between periodic and continuous monitoring as required, based on a fast detection algorithm for precipitation particles (FDAP). FDAP uses a background model and thresholding strategy to segment precipitation particles. In particular, it applies an area rule and a neighborhood rule to eliminate the influence of noise and external interference (e.g., flying insects) on the field observations. We describe precipitation particles using a composite representation that includes geometric features, Fourier descriptors, and Hu moment invariants, and also adopt gradient boosting trees to classify the type of precipitation. The experimental evaluation of POTR on data sets collected on-site in Beijing from August 2014 to February 2015 shows that FDAP has an accuracy of more than 96% and that type recognition has an accuracy more than 90%.