Fourier Descriptor Method Research Articles

Hybrid Fourier Descriptor Naïve Bayes dan CNN pada Klasifikasi Daun Herbal

Plants are vital to human life on earth, and the leaves and their whole parts have many benefits. These parts of the plant can help distinguish between different species. The leaf identification can be performed at any time, while the other parts of the plants can only be identified at a certain time. The study aims to classify two types of herbs i.e. saur-opus androgynous and moringa oleifera, implementing the Fourier Descriptor method to extract the shape and texture features. In the process of classification using the Naïve Bayes method with three types of nuclei (Gaussian, Bernoulli, and Multinomial) and a Convolutional Neural Network. The testing process was carried out using two scenarios, dark and light, where each scenario consisted of 240 images for a total of 480 images divided into 20% of the data testing and 80% of the training data. The Fourier Descriptor-Bernoulli Naive Bayes method gives the lowest accuracy in both light and dark scenarios, at 46% and 52%, respectively. As for the classification of herbal leaves using a combination of the Fourier Descriptor-Convolutional Neural Network method, it is recommended to be used in light image scenarios and Fourier Descriptor-Gaussian Naive Bayes in the dark scenarios because it is able to detect herbal leaf types with 100% accuracy.

Gabor Descriptor for Representation of Spatial Feature

New spatial feature descriptor based on Gabor wavelet function is proposed. The proposed method is compared to Fourier descriptor. The experimental results with Advanced Earth Observing Satellite: ADEOS / Advanced Visible and Near-Infrared Radiometer: AVNIR image show an effectiveness of the proposed method. It is found that the restored image quality, in terms of root mean square error between the original and the restored images depends on the support length of the mother wavelet and is much better than that with the conventional Fourier descriptor method for spatial feature description.

A Novel UKF-RBF Method Based on Adaptive Noise Factor for Fault Diagnosis in Pumping Unit

Fault detection and diagnosis in the pumping unit is a challenging industrial problem for the system that exhibits nonlinearity, coupled parameters, and time-varying noise. This paper proposes a novel combined unscented Kalman filter (UKF) and radial basis function (RBF) method based on an adaptive noise factor for fault diagnosis in the pumping unit. First, to reduce computation and complexity of the diagnosis model, the Fourier descriptor method based on an approximate polygon is presented to extract the features of the indicator diagram. RBF neural network is adopted to establish the fault diagnosis model based on indicator diagram data and production data. In particular, UKF is used to train the weights ( $w_{m,l}$ ), the center ( $c_{m}$ ), and the width ( $b_{m}$ ) of the RBF model. Furthermore, the adaptive noise factor method is proposed to address the adaptive filtering issue in the fault diagnosis model. The proposed method is applied to the pumping unit system, and experimental results show the effectiveness and favorable recognition rate in classifying multiple faults.

Facilitating the Automatic Characterisation, Classification and Description of Biological Images with the VisionBioShape Package for R

Here, we present the VisioBioshapeR package for R [R Core, 2014]. This new library is a comprehensive, multifunctional toolbox designed to automatically analyse biological images. The package extends other common libraries (Momocs, ShapeR) used for biological shape analysis by allowing the user to extract closed contour outlines automatically from reading binary images. Current functionalities of VisioBioshapeR include: random extraction of image coordinates, analysis of the shape of a biological image by the elliptic Fourier descriptor (EFD) method, extraction of an image characteristic vector using multivariate principal component analysis (PCA) and geometrical analysis. The image vector of characteristics can be directly exported to a wide range of statistical packages in R and can be used to perform classification or other types of analysis in order to sort new images into classes. The package could prove useful in studies of any two-dimensional images and is presented with three examples of its application in ecology. The library is useful when multiple images are processed at a time and we wish to automate their analysis for example for recognition of images from patterns.

EDIBLE BIRD NEST SHAPE INSPECTION USING FOURIER DESCRIPTOR (FD) AND FARTHEST FOURIER POINT SIGNATURE (FFPS) METHOD

Swiftlets are birds contained within the four genera Aerodramus, Hydrochous, Schoutedenapus and Collocalia. To date, the bird nest grading is based on weight, shape and size. Current inspection and grading for raw, edible bird nest were performed visually by expert panels. This conventional method is relying more on human judgments and often biased. A novel hybrid method from Fourier Descriptor (FD) method and Farthest Fourier Point Signature (FFPS) was developed using Charge Coupled Device (CCD) image data to grade bird nest by its shape and size. From the result, the hybrid method was able to differentiate different shape such as super AAA, super and corner grade depending on the Swiftlet species and geographical origin. The Wilks' lambda analysis was invoked to transform and compress the data set comprising of a large number of interconnected variables to a reduced set of varieties. Overall, the vision system was able to correctly classify 92.6 % of the super AAA, super and Corner shaped grades using the combined FD and FFPS features.

Comparison of leaf surface roughness analysis methods by sensitivity to noise analysis

Surface roughness is of great interest in agricultural spraying because it is used to characterise leaf surface wettability to predict the behaviour of droplets on a leaf surface. In recent years, the use of texture analysis to estimate surface roughness has emerged. In this paper we propose to estimate leaf surface roughness by using an optimisation of the Generalized Fourier Descriptors method. This approach is then compared with two other standard methods in the literature, one based on grey level intensity variation and the other on wavelet decomposition. Since roughness has many definitions and each method is calculated differently, we propose a new approach to compare the results based on the sensitivity of each method according to surface roughness variations. These variations were introduced by adding different kinds of noise to the image. Gaussian and salt & pepper noise are added to simulate rapid changes and peak impulses on the surface topography, whereas a Structural noise (sinusoidal signal) is added to simulate depth on the surface topography. The novelty of this contribution is the use of a new approach and procedure for agronomic application (leaf surface roughness). The results obtained are expected to be used to characterise the adhesion mechanisms of liquid droplets on a leaf surface.

Shape and weight grading of mangoes using visible imaging

This paper presents the work on the use of visible imaging as a tool in grading the mangoes. A Fourier-descriptor method was applied on mango images acquired by a CCD camera, to grade the fruits by their shapes. The method was able to correctly classify 98.3% using DA and 100% using SVM. It is also possible to estimate the weight of the mangoes from their images by applying the Cylinder approximation analysis method. The scatter plot between the estimated and actual values of the weight shows high correlation, with R2 equal to 94.0%. The high prediction accuracy obtained shows that this simple formula is adequate for the prediction of fruit weight and volume (measured volume using the cylinder method). The correlation formula derived based on the collected data is determined as w=2.256V−157.7 where w is estimated weight in grams and V is estimated volume. Overall result for weight grading using our proposed method yields 95% accuracy.

Comparison Contour Extraction Based on Layered Structure and Fourier Descriptor on Image Retrieval

In this paper, a new content-based image retrieval technique using shape feature is proposed. A shape features extracted by layered structure representation has been implemented. The approach is extract feature shape by measuring the distance between centroid (center) and boundaries of the object that can capture multiple boundaries in the same angle, an object shape that has some points with the same angle.Once an input taking into account, the method will search most related image to the input. The correlation between input and output has been defined by specific role. Firstly the input image has to be converted from RGB image to Grayscale image and then follow by edge detection process. After edge detection process the boundary object will be obtained and then calculate the distance between the center of an object and the boundary of an object and put it in the feature vector and if there is another boundary on the same angle then put it in the different feature vector with a different layer. The experiment result on the plankton dataset shows that the proposed method better than other conventional Fourier descriptor method.

Neural Network based Approach for Recognition of Text Images

Handwritten character recognition is a difficult problem due to the great variations of writing styles, different size of the characters. Multiple types of handwriting styles from different persons are considered in this work. An image with higher resolution will certainly take much longer time to compute than a lower resolution image. In the practical image acquisition systems and conditions, shape distortion is common processes because different people’s handwriting has different shape of characters. The process of recognizing character recognition in this work has been divided into 2 phases. In the first phase, Image preprocessing is done in which image is firstly converted into binary form based on some threshold value obtained through Otsu’s method. After that removal of noise is done using median filter. After that feature extraction takes place that is done here through Fourier descriptor method using Fourier transform and correlation between template made through training data and test data is obtained. A multilayer feed forward neural network is created and trained through Back Propagation algorithm. After the training, testing is done to match the pattern with test data. Results for various convergence objective of neural network are obtained and analyzed.

A method of building fourier descriptors for partial shapes and some properties.

A method of building fourier descriptors for partial shapes and some properties.

Hand Gesture Recognition using Optical Flow Field Segmentation and Boundary Complexity Comparison based on Hidden Markov Models

In this paper, we will present a method to detect human hand and recognize hand gesture. For detecting the hand region, we use the feature of human skin color and hand feature (with boundary complexity) to detect the hand region from the input image; and use algorithm of optical flow to track the hand movement. Hand gesture recognition is composed of two parts: 1. Posture recognition and 2. Motion recognition, for describing the hand posture feature, we employ the Fourier descriptor method because it's rotation invariant. And we employ PCA method to extract the feature among gesture frames sequences. The HMM method will finally be used to recognize these feature to make a final decision of a hand gesture. Through the experiment, we can see that our proposed method can achieve 99% recognition rate at environment with simple background and no face region together, and reduce to 89.5% at the environment with complex background and with face region. These results can illustrate that the proposed algorithm can be applied as a production.

Modelling nasal airflow using a Fourier descriptor representation of geometry

AbstractProcedures capable of providing both compact representations and rational simplifications of complex anatomical flow conduits are essential to explore how form and function are related in the respiratory, cardiovascular and other physiological flow systems. This study focuses on flow in the human nasal cavity. Methods to derive the cavity wall boundary from medical images are first outlined. Anisotropic smoothing of the boundary surface is shown to provide less geometric distortion in regions of high curvature, such as at the ends of the narrow nasal passages. A reversible decomposition of the surface into a stack of planar contours is then effected using an implicit function formulation. Fourier descriptors provide a continuous representation of each contour as a modal expansion and permit simplification of the geometry by retaining only dominant modes via filtering.Computations of the steady inspiratory flow field are performed for replica and reduced geometries, where reduced geometry is derived by retaining only the first 15 modes in the expansion of each slice contour. The overall pressure drop and integrated wall shear are shown to be virtually unaffected by simplification. More sensitive measures, such as the Lagrangian particle trajectories and residence time distributions, show slight changes as discussed.Comparison of the Fourier descriptor method applied to three different patient data sets indicates the potential of the technique as a means to characterize complex flow conduit geometry, and the scope for further work is outlined. Copyright © 2008 John Wiley & Sons, Ltd.

Affine Invariant Contour Descriptors Using Independent Component Analysis and Dyadic Wavelet Transform

The paper presents a novel technique for affine invariant feature extraction with the view of object recognition based on parameterized contour. The proposed technique first normalizes an input image by removing the affine deformations using independent component analysis which also reduces the noise introduced during contour parameterization. Then four invariant functionals are constructed using the restored object contour, dyadic wavelet transform and conics in the context of wavelets. Experimental results are conducted using three different standard datasets to confirm the validity of the proposed technique. Beside this the error rates obtained in terms of invariant stability are significantly lower when compared to other wavelet based invariants. Also the proposed invariants exhibit higher feature disparity than the method of Fourier descriptors.

Shape recognition based on neural networks trained by differential evolution algorithm

In this paper a new method for recognition of 2D occluded shapes based on neural networks using generalized differential evolution training algorithm is proposed. Firstly, a generalization strategy of differential evolution algorithm is introduced. And this global optimization algorithm is applied to train the multilayer perceptron neural networks. The proposed algorithms are evaluated through a plant species identification task involving 25 plant species. For this practical problem, a multiscale Fourier descriptors (MFDs) method is applied to the plant images to extract shape features. Finally, the experimental results show that our proposed GDE training method is feasible and efficient for large-scale shape recognition problem. Moreover, the experimental results illustrated that the GDE training algorithm combined with gradient-based training algorithms will achieve better convergence performance.

Shape based leaf image retrieval

The authors present an efficient two-stage approach for leaf image retrieval by using simple shape features including centroid–contour distance (CCD) curve, eccentricity and angle code histogram (ACH). In the first stage, the images that are dissimilar with the query image will be first filtered out by using eccentricity to reduce the search space, and fine retrieval will follow by using all three sets of features in the reduced search space in the second stage. Different from eccentricity and ACH, the CCD curve is neither scaling-invariant nor rotation-invariant. Therefore, normalisation is required for the CCD curve to achieve scaling invariance, and starting point location is required to achieve rotation invariance with the similarity measure of CCD curves. A thinning-based method is proposed to locate starting points of leaf image contours, so that the approach used is more computationally efficient. Actually, the method can benefit other shape representations that are sensitive to starting points by reducing the matching time in image recognition and retrieval. Experimental results on 1400 leaf images from 140 plants show that the proposed approach can achieve a better retrieval performance than both the curvature scale space (CSS) method and the modified Fourier descriptor (MFD) method. In addition, the two-stage approach can achieve a performance comparable to an exhaustive search, but with a much reduced computational complexity.

An Approach for Image Classification using Backpropagation Scheme

In this paper certain simple procedures are presented for image classification using a class of Neural Network (NN) trained by backpropagation algorithm with minimum number of failures. The representation of the images is obtained through predominant Fourier Descriptor Method (FDM). To quicken the network response (Error vs Epoch) with minimum number of Oscillations, a slope parameter is utilized in the bipolar sigmoidal activation function in the hidden and output layers. The above scheme is applied to a Feed-Forward Neural Network (FFNN) for image classification. The performance of the network in terms of local minima, learning speed is evaluated for the choices of slope parameter, learning rate, momentum factor and compared with the traditional BP Algorithm.

Region-based shape representation and similarity measure suitable for content-based image retrieval

A region-based approach to shape representation and similarity measure is presented. The shape representation is invariant to translation, scale and rotation. The similarity measure conforms to human similarity perception, i.e., perceptually similar shapes have high similarity measure. An experimental shape retrieval system has been developed and its performance has been studied. The shape retrieval performance of the proposed approach is better than that of the more established Fourier descriptor method.

Application of elliptic Fourier descriptors to symmetry detection under parallel projection

In this paper, the method of elliptic Fourier descriptors using arc length parameterization is applied to tackle the problem of detection and recovery of symmetry under parallel projection. A simple and fast iteration algorithm together with the invariants of symmetry provides sufficient information for the detection and recovery of symmetry under parallel projection. The proposed method has been extensively tested using symmetric figures under different parallel projections. Simulation results of the algorithm are presented. The extension of this method for planar object recognition under parallel projection is also addressed. >

Quantification of function regulator therapy using elliptical Fourier functions.

The application of various functional appliances to correct antero-posterior skeletal discrepancies in the treatment of Class III malocclusion has received increased attention. In this context, elliptical Fourier functions (EFF) are particularly useful for eliciting boundary-outline information. This longitudinal study was undertaken to visually and numerically describe the shape alterations arising as a response to treatment with the FR-3 appliance of Fränkel. The subjects chosen were Caucasians (n = 14) from the Glasgow Dental Hospital and School who presented with Class III incisor relationships. Lateral cephalometric head-films were available prior to the initiation of treatment, at the end of active treatment, and at least one year out of retention. Maxillary and mandibular boundary outlines were carefully traced. A set of points, 42 on the maxilla, and 78 on the mandible, were precisely located. These points were then digitized and used to compute EFF's with 20 harmonics. These equations were then standardized for size. Subsequently, 102 maxillary and 150 mandibular closely spaced points on the outline were calculated from the EFF's. After superimposition on the centroid, expected vectors were then computed to these points from the centroid. Statistically significant dento-alveolar shape changes could be discerned with treatment. These involved a localized anterior repositioning of the maxillary incisal aspect with a concomitant posterior displacement of the mandibular incisal aspect. There was no evidence of shape changes in the basal areas of either the maxilla or mandible. This method of fitting Fourier descriptors, specifically EFF's, provides a precise numerical and visual description of the changes arising from functional regulator therapy.

ROBUST MORPHOLOGICALLY CONTINUOUS FOURIER DESCRIPTORS I: PROJECTION-GENERATED DESCRIPTORS

By generating Fourier descriptors based upon the waveform induced by a pattern's geometric projection, a number of classic difficulties with the Fourier-descriptor methodology are mitigated. Not only are the descriptors invariant with respect to scale, translation, and rotation (as is usually the case), they are also continuous in the Hausdorff metric and robust with respect to both point noise and occlusion. An additional advantage is that they can be computed relative to a thresholded image without first finding an edge, thereby avoiding the difficulties typically present in thinning and orientation determination. The present paper discusses the method of projection-generated Fourier descriptors, as well as a study of the sensitivity to point noise. A companion paper will present the morphological properties and the effect of pattern occlusion.