International Journal of Image, Graphics and Signal Processing @ijigsp
Статьи журнала - International Journal of Image, Graphics and Signal Processing
Все статьи: 1146
Статья научная
This paper addresses the problem of detecting the partially-correlated χ2 fluctuating targets with two and four degrees of freedom. It presents the performance analysis, in its exact form, of GTM-CFAR processor when the operating environment is contaminated with extraneous targets and the radar receiver post-detection integrates M pulses of exponentially correlated targets. Mathematical formulas for the detection and false alarm probabilities are derived, in the absence as well as in the presence of spurious targets which are fluctuating in accordance with the so-called moderately fluctuating χ2 targets. A thorough performance assessment by several numerical examples, which has considered the role that each parameter can play in the processor performance, is also given. The results show that the processor performance improves, for weak SNR of the primary target, as the correlation coefficient ρs increases and this occurs either in the absence or in the presence of outlying targets. As the strength of the target return increases, the processor tends to invert this behavior. The SWI & SWII and SWIII & SWIV models enclose the correlated target cases when the target correlation follows χ2 fluctuation models with two and four degrees of freedom, respectively, and this behavior is common for all GTM based detectors.
Бесплатно
Patch based image inpainting technique using adaptive patch size and sequencing of priority terms
Статья научная
Image Inpainting is a system used to fill lost information in an image in a visually believable manner so that it seems original to the human eye. Several algorithms are developed in the past which tend to blur the inpainted image. In this paper, we present an algorithm that improves the performance of patch based image inpainting by using adaptive patch size and sequencing of the priority terms. The patch width (wxw) is made adaptive (proportional) to the area of the damaged region and inversely proportional to standard deviation of the known values in the patch around point of highest priority. If the neighbourhood region is a smooth region then standard deviation is small therefore large patch size is used and if standard deviation is large patch size is small. The algorithm is tested for various input images and compared with some standard algorithm to evaluate its performance. Results show that the time required for inpainting is drastically reduced while the quality factor is maintained equivalent to the existing techniques.
Бесплатно
Pattern Recognition: Invariance Learning in Convolutional Auto Encoder Network
Статья научная
The ability of the human visual processing system to accommodate and retain clear understanding or identification of patterns irrespective of their orientations is quite remarkable. Conversely, pattern invariance, a common problem in intelligent recognition systems is not one that can be overemphasized; obviously, one's definition of an intelligent system broadens considering the large variability with which the same patterns can occur. This research investigates and reviews the performance of convolutional networks, and its variant, convolutional auto encoder networks when tasked with recognition problems considering invariances such as translation, rotation, and scale. While, various patterns can be used to validate this query, handwritten Yoruba vowel characters have been used in this research. Databases of images containing patterns with constraints of interest are collected, processed, and used to train and simulate the designed networks. We provide extensive architectural and learning paradigms review of the considered networks, in view of how built-in invariance is learned. Lastly, we provide a comparative analysis of achieved error rates against back propagation neural networks, denoising auto encoder, stacked denoising auto encoder, and deep belief network.
Бесплатно
Pattern averaging technique for facial expression recognition using support vector machines
Статья научная
Facial expression is one of the nonverbal communication methods of identifying an emotional state of a human being. Due to its crucial importance in Human-Robot interaction, facial expression recognition (FER) is in the limelight of recent research activities. Most of the studies consider the whole expression images in their analysis, and it has several has several drawbacks concerning illumination, orientation, texture, zoom level, time and space complexity. In this paper, a novel feature extraction technique called the pattern averaging is studied on whole image data using reduction in the dimension of the image by averaging the neighboring pixels. The study is found to give better results on standard datasets using support vector machine classifier.
Бесплатно
Pavement Crack Detection Using Spectral Clustering Method
Статья научная
Pavement crack detection plays an important role in pavement maintaining and management, nowadays, which could be performed through remote image analysis. Thus, edges of pavement crack should be extracted in advance; in general, traditional edge detection methods don’t consider phase information and the spatial relationship between the adjacent image areas to extract the edges. To overcome the deficiency of the traditional approaches, this paper proposes a pavement crack detection algorithm based on spectral clustering method. Firstly, a measure of similarity between pairs of pixels is taken into account through orientation energy. Then, spatial relationship is needed to find regions where similarity between pixels in a given region is high and similarity between pixels in different regions is low. After that, crack edge detection is completed with spectral clustering method. The presented method has been run on some real life images of pavement crack, experimental results display that the crack detection method of this paper could obtain ideal result.
Бесплатно
Pedestrian Detection in Thermal Images Using Deep Saliency Map and Instance Segmentation
Статья научная
Pedestrian detection is an established instance of computer vision task. Pedestrian detection from the color images has achieved robust performance but in the night time or in bad light conditions it has low detection accuracy. Thermal images are used for detecting people at night time, foggy weather or in bad lighting situations when color images have a lower vision. But in the daytime where the surroundings are warm or warmer than pedestrians then the thermal image has lower accuracy. Hence thermal and color image pair can be a solution but it is expensive to capture color-thermal pair and misaligned imagery can cause low detection accuracy. We proposed a network that achieved better accuracy by extending the prior works which introduced the use of the saliency map in pedestrian detection tasks from the thermal images into instance-level segmentation. We worked on a subdivision of KAIST Multispectral Pedestrian Detection Dataset [8] which has pixel-level annotations. We have trained Mask-RCNN for pedestrian detection task and report the added effect of saliency maps generated using PiCA-Net. We have achieved an accuracy of 88.14% over day and 91.84% over night images. So, our model has reduced the miss rate by 24.1% and 23% over the existing state-of-the-art method in day and night images.
Бесплатно
Статья научная
Image contrast is very important visual characteristics that will considerably improve the appearance of the image. In this paper image contrast is to be enhanced optimally to accurately portray all the data in the image using nature inspired meta-heuristic algorithms. Algorithms have been devised and proposed to enhance the contrast of low contrast images in this work. Poor image contrast caused by a low-quality capturing device, biased user experience, and an unsuitable environment setting during image capture is the main problem encountered during the image enhancement process. Histogram Equalization (HE), a frequently used technique for contrast enhancement, typically produces images with unwanted artifacts, an unnatural appearance, and washed-out appearances. The degree of enhancement is beyond the control of the global HE. The quality of an image is crucial for human comprehension, making image contrast enhancement (ICE) a crucial pre-processing stage in image processing and analysis. In the current study, the Pelican Optimization Algorithm, a contemporary meta-heuristic (MH) algorithm influenced by nature, is used as the foundation for the grayscale image contrast enhancement (GICE) approach (POA). The comparison of proposed method with existing contrast enhancement techniques has been done on the basis of standard image quality metrics. The proposed algorithm performance on standard test image and Kodak dataset demonstrates that total image contrast and information provided in the image are both greatly improved by the suggested POA-based image enhancement technique.
Бесплатно
Статья научная
A human eye can detect a face in an image whether it is in a digital image or also in some video. The same thing is highly challenging for a machine. There are lots of algorithms available to detect human face. In this paper, a technique has been made to detect and locate the position of human faces in digital images. This approach has two steps. First, training the artificial neural network using Levenberg–Marquardt training algorithm and then the proposed algorithm has been used to detect and locate the position of the human faces from digital image. The proposed algorithm has been implemented for six color spaces which are RGB, YES, YUV, YCbCr, YIQ and CMY for each of the image formats bmp, jpeg, gif, tiff and png. For each color space training has been made for the image formats bmp, jpeg, gif, tiff and png. Finally, one color space and particular image format has been selected for face detection and location in digital image based on the performance and accuracy.
Бесплатно
Статья научная
Object Tracking is becoming increasingly important in areas of computer vision, surveillance, image processing and artificial intelligence. The advent of high powered computers and the increasing need of video analysis has generated a great deal of interest in object tracking algorithms and its applications. This said it becomes even more important to evaluate these algorithms to quantify their performance. In this paper, we have implemented three algorithms namely Alpha Beta filter, Kalman filter and Meanshift to track an object in a video sequence and compared their tracking performance based on various parameters in normal and noisy conditions. The proposed parameters employed are error plots in position and velocity of the object, Root mean square error, object tracking error, tracking rate and time taken to track the object. The goal is to illustrate practically the performance of each algorithm under such conditions quantitatively and identify the algorithm that performs the best.
Бесплатно
Performance Analysis of Fingerprint Denoising Using Stationary Wavelet Transform
Статья научная
Finger print is the finest and cheapest recognition system because of its easy extraction of unique features like bifurcation and termination. But the quality of fingerprint data are easily degraded by dryness of skin, wet, wound and other types of noises. Hence, denoising of fingerprint image is vital step for automatic fingerprint recognition system. In the proposed paper the removal of noise from fingerprint images by using stationary wavelet transform and adaptive thresholding method is analysed. The proposed algorithm is developed using MATLAB (R2010b) and tested in the fingerprint images collected from FVC2004 database and R303A optical scanner. The performance of the method is analysed by calculating the quality metrics like Peak Signal to Noise Ratio, Universal Quality Index , Structure Similarity and Multi-Scale Structure Similarity (MS-SSIM). The quality of fingerprint image after noise removal using proposed analysis confirms the suggested method is better than the conventional techniques.
Бесплатно
Performance Analysis of Non-Linear Equalizer in MIMO System for Vehicular Channel
Статья научная
All wireless technologies face the challenges of multipath signal fading, attenuation delay and phase delay which led to the interference between users and there is the possibility of limited spectrum. Linear and Non-Linear receiver is used to combat the effect of multipath signal fading and delay. The linear receiver gives best result in case of static environment but in case of dynamic environmental condition it fails to give better results and hence in order to improve the system performance non-linear receiver is used in dynamic environment condition. As a dynamic channel, Vehicular Channel model is considered because there is growing interest in vehicular networking and it is also a challenging channel model because of the complexity of the environment, and rapid variation in channel conditions. This paper studies the comparison between Zero Forcing (ZF) and Minimum Mean Square Error (MMSE) receiver in the Vehicular Channel. A comparative study between linear equalizer and non-linear equalizer in the Vehicular Channel is done and analyze the effect of the varying modulation and antenna configuration on the performance.
Бесплатно
Performance Analysis of Texture Image Classification Using Wavelet Feature
Статья научная
This paper compares the performance of various classifiers for multi class image classification. Where the features are extracted by the proposed algorithm in using Haar wavelet coefficient. The wavelet features are extracted from original texture images and corresponding complementary images. As it is really very difficult to decide which classifier would show better performance for multi class image classification. Hence, this work is an analytical study of performance of various classifiers for the single multiclass classification problem. In this work fifteen textures are taken for classification using Feed Forward Neural Network, Naïve Bays Classifier, K-nearest neighbor Classifier and Cascaded Neural Network.
Бесплатно
Performance Analysis of Various Image Feature Extractor Filters for Pothole Anomaly Classification
Статья научная
Machine learning (ML) classifiers have lately gained traction in the realm of intelligent transportation systems as a means of enhancing road navigation while also assisting and increasing automotive user safety and comfort. The feature extraction stage, which defines the performance accuracy of the ML classifier, is critical to the success of any ML classifiers used. Nonetheless, the efficacy of various ML feature extractor filters on image data of road surface conditions obtained in a variety of illumination settings is uncertain. Thus, an examination of eight different feature extractor filters, namely Auto colour, Binary filter, Edge Detection, Fuzzy Color Texture Histogram Filter (FCTH), J-PEG Color, Gabor filter, Pyramid of Gradients (PHOG), and Simple Color, for extracting pothole anomalies feature from road surface conditions image data acquired under three environmental scenarios, namely bright, hazy, and dim conditions, prior classification using J48, JRip, and Random Forest ML models. According to the results of the experiments, the auto colour image filter is better suitable for extracting features for categorizing road surface conditions image data in bright light circumstances, with an average classification accuracy of roughly 96%. However, with a classification accuracy of around 74%, the edge detection filter is best suited for extracting features for the classification of road surface conditions image data captured in hazy light circumstances. The autocolor filter, on the other hand, has an accuracy of roughly 87% when it comes to classifying potholes in low-light conditions. These findings are crucial in the selection of feature extraction filters for use by ML classifiers in the development of a robust autonomous pothole detection and classification system for improved navigation on anomalous roads and possible integration into self-driving cars.
Бесплатно
Статья научная
Tropical cyclones, considered extreme weather events, can cause significant damage to coastal areas, impacting millions of people and animals while also posing the risk of substantial economic losses. Traditionally, the Dvorak technique has been employed to assess the intensity of these cyclones, involving the visual analysis of satellite data to evaluate the storm’s cloud patterns and strength. In recent years, various studies have explored the use of deep learning (DL) and machine learning (ML) techniques to estimate tropical cyclone intensity. However, there is a lack of research providing a comparative analysis that integrates both ML and DL approaches for the estimation of tropical cyclone intensity. This study looks into the use of ML and DL techniques to estimate the strength of tropical cyclones. On diverse datasets and satellite imagery, we study the usage of convolutional neural networks (CNN, VGG16, DenseNet), recurrent neural networks (LSTM), and other machine learning methods (XGBoost, CatBoost, SVM, DT). Our findings suggest that both ML and DL methods have substantial promise for improving tropical cyclone intensity estimation accuracy; however, in our case study, DL algorithms outperformed ML algorithms. This study investigates the utilization of ML and DL techniques in assessing the strength of tropical cyclones. Employing various datasets and satellite imagery, we examine the performance of convolutional neural networks (CNNs such as VGG16 and DenseNet), recurrent neural networks (LSTM), and other ML methods (XGBoost, CatBoost, SVM, DT). Our results indicate that both ML and DL approaches show significant promise in enhancing the accuracy of tropical cyclone intensity estimation. Nevertheless, in our specific case study, DL algorithms demonstrated superior performance compared to ML algorithms.
Бесплатно
Статья научная
In this paper image compression using hybrid wavelet transform is proposed. Hybrid wavelet transform matrix is formed using two component orthogonal transforms. One is base transform which contributes to global features of an image and another transform contributes to local features. Here base transform is varied to observe its effect on image quality at different compression ratios. Different transforms like Discrete Kekre Transform (DKT), Walsh, Real-DFT, Sine, Hartley and Slant transform are chosen as base transforms. They are combined with Discrete Cosine Transform (DCT) that contributes to local features of an image. Sizes of component orthogonal transforms are varied as 16-16, 32-8 and 64-4 to generate hybrid wavelet transform of size 256x256. Results of different combinations are compared and it has been observed that, DKT as a base transform combined with DCT gives better results for size 16x16 of both component transforms.
Бесплатно
Статья научная
A hybrid watermarking technique using Singular value Decomposition with orthogonal transforms like DCT, Haar, Walsh, Real Fourier Transform and Kekre transform is proposed in this paper. Later, SVD is combined with wavelet transforms generated from these orthogonal transforms. Singular values of watermark are embedded in middle frequency band of column/row transform of host image. Before embedding, Singular values are scaled with suitable scaling factor and are sorted. Column/row transform reduces the computational complexity to half and properties of singular value decomposition and transforms add to robustness. Behaviour of proposed method is evaluated against various attacks like compression, cropping, resizing, and noise addition. For majority of attacks wavelet transforms prove to be more robust than corresponding orthogonal transform from which it is generated.
Бесплатно
Performance Evaluation and Comparative Analysis of Different Filters for Noise Reduction
Статья научная
The quality of microscopic images is generally degraded during the image acquisition by quantizing noise, electrical noise, light illumination etc. Noise reduction is considered as a very important preprocessing step as the quality of the images can determine the accuracy of the results. The work done focuses on the noise reduction using different filters on the different types of noises applied on the common digital images and specifically the Leukemia images. 40 images were taken for the comparison purpose; 20 digital images and 20 Leukemia images of different types of Leukemia. The qualitative as well as quantitative analysis of the performance of the filters on the different noises is done. For the quantitative analysis the parameters used for the evaluation of the images are MSE, PSNR and CoC. For the qualitative analysis visual analysis in terms of quality is also done using the resultant images and their histograms. Simulation has been done in Matlab 11b. From the test cases it has been observed that Adaptive Filter produces good results on Salt and Pepper, Speckle and Gaussian noise in case of the digital images. Whereas in case of Leukemia images results of Median Filter are best for the Gaussian, Poisson and Speckle noise corrupted images.
Бесплатно
Performance Evaluation of Image Fusion Algorithms for Underwater Images-A study based on PCA and DWT
Статья научная
In this paper, a comparative study between two image fusion algorithm based on PCA and DWT is carried out in underwater image domain. Underwater image fusion is emerged as one of the main image fusion area, here two or more images will be fused by retaining the most desirable characteristics of each underwater images. The DWT technique is used to decompose the input image into four frequency sub bands and the low-low sub band images will be considered in fusion processing. In PCA method significant eigen values will be considered in fusion process to retain the important characteristics of the input images. The results acquired from both experiments are tabulated and compared by considering the statistical measures such as Peak Signal to Noise Ratio (PSNR), Mean Square Error (MSE) and Entropy. Results shows that underwater image fusion based on DWT outperforms the PCA based method.
Бесплатно
Статья научная
The present paper aims at performance evaluation of Doubly Truncated Generalized Laplace Mixture Model and Hierarchical clustering (DTGLMM-H) for image analysis concerned to various practical applications like security, surveillance, medical diagnostics and other areas. Among the many algorithms designed and developed for image segmentation the dominance of Gaussian Mixture Model (GMM) has been predominant which has the major drawback of suiting to a particular kind of data. Therefore the present work aims at development of DTGLMM-H algorithm which can be suitable for wide variety of applications and data. Performance evaluation of the developed algorithm has been done through various measures like Probabilistic Rand index (PRI), Global Consistency Error (GCE) and Variation of Information (VOI). During the current work case studies for various different images having pixel intensities has been carried out and the obtained results indicate the superiority of the developed algorithm for improved image segmentation.
Бесплатно
Статья научная
Super resolution (SR) images play an important role in Image processing applications. Spatial resolution is the key parameter in many applications of image processing. Super resolution images can be used to improve the spatial resolution. In this paper a new SR image reconstruction algorithm is proposed using Integer wavelet transform (IWT) and Binary plane technique (BPT). The proposed method is analyzed in different color space transforms such as CIELAB, YCbCr and RGB. In this paper we compared PSNR, ISNR, Blocking effect and Homogeneity with different colour images in RGB, YCbCr and CIELAB domains. Qualitative analysis shows that the proposed method in CIELAB color space transforms has better performance.
Бесплатно
