Статьи журнала - International Journal of Image, Graphics and Signal Processing
Все статьи: 1092
Image Recognition Using Machine Learning with the Aid of MLR
Статья научная
In this paper, we use three machine learning techniques: Linear Discriminant Analysis (LDA) along different Eigen vectors of an image, Fuzzy Inference System (FIS) and Fuzzy c-mean clustering (FCM) to recognize objects and human face. Again, Fuzzy c-mean clustering is combined with multiple linear regression (MLR) to reduce the four-dimensional variable into two dimensional variables to get the influence of all variables on the scatterplot. To keep the outlier within narrow range, the MLR is again applied in logistic regression. Individual method is found suitable for particular type of object recognition but does not reveal standard range of recognition for all types of objects. For example, LDA along Eigen vector provides high accuracy of detection for human face recognition but very poor performance is found against discrete objects like chair, butterfly etc. The FCM and FIS are found to provide moderate result in all kinds of object detection but combination of three methods of the paper provide expected result with low process time compared to deep leaning neural network.
Бесплатно
Image Recognition by Using the Progressive Wavelet Correlation
Статья научная
An algorithm for image recognition and retrieval of image from image collection is developed. Basis of the algorithm is the progressive wavelet correlation. The recognition consists of three incremental steps, each of them quadruples the number of correlation points. The process can be aborted at any stage if the intermediate results indicate that the correlation will not result in a match. The final result is the recognition and retrieval of the required image, if exists in the image collection. Instructions for the choice of correlation threshold value for obtaining desired results are defined. We perform a series of image search experiments that cover the following scenarios: the given image is present in the database; copies of the given image are present but with different names; similar (but not identical) images are present; and the given image is not present. Experiments are performed with data bases up to 1000 images, using the Oracle database and the Matlab component Database Toolbox for operations with databases.
Бесплатно
Image Resolution Enhancement by Using Stationary and Discrete Wavelet Decomposition
Статья научная
This work proposed an image resolution enhancement technique which is based on the interpolation of the high frequency subbands obtained by DWT. The proposed technique uses DWT to decompose an image into different subbands, and then the high frequency subband images have been interpolated. The interpolated high frequency subband coefficients have been corrected by using the high frequency subbands achieved by SWT of the input image. An original image is interpolated with half of the interpolation factor used for interpolation the high frequency subbands. Afterwards all these images have been combined using IDWT to generate a super resolved imaged. The proposed technique has been tested on well-known benchmark images, where their PSNR, Mean Square Error and Entropy results show the superiority of proposed technique over the conventional and state-of-art image resolution enhancement techniques.
Бесплатно
Статья научная
This research is focusing on ornamental leaf with dual functionalities, which are ornamental and medicinal functionalities. However, only few people know about the medicinal functionality of this plant. In Indonesia, this plant is also easy to find because mostly cultivates in front of the house. If its medicinal function and that easiness are taken into consideration, this leaf should be an option towards the full chemical-based medicines. This image retrieval system utilizes color, shape, and texture features from leaf images. HSV-based color histogram, Zernike complex moments, and Dyadic wavelet transformation are the color, shape, and texture features extractor methods, respectively. We also implement the Bayesian automatic weighting formula instead of assignment of static weighting factor. From the results, this proposed method is very powerful from any rotation, lighting, and perspective changes.
Бесплатно
Image Retrieval by Utilizing Structural Connections within an Image
Статья научная
Content-based image retrieval (CBIR) is broadly applicable for searching digital images from a gigantic database. Images are retrieved by their primitive visual contents such as color, texture, shape, and spatial layout. The approach presented in this paper utilizes structural connections within an image by integrating textured color descriptors and structure descriptors to retrieve semantically significant images. The retrieval results were obtained by applying the HSV histogram, color coherence vector, and local binary pattern histogram to the standard database of Wang et al., which has 1000 images of 10 different semantic categories. Euclidean distance was used to find the similarity between the query image and database images. This method was evaluated against different methods based on edge histogram descriptors, color structure descriptors, color moments, the color histogram, the HSV histogram, Tamura features, edge descriptors, geometrical shape attributes, and statistical properties such as mean, variance, skewness, and kurtosis. Retrieval results obtained using the proposed methods demonstrated a significant improvement in the average precision (73.8% and 73.1%) compared with those obtained using other existing retrieval methods.
Бесплатно
Image Segmentation Method for Identifying Convective and Stratiform Rain using MSG SEVIRI Data
Статья научная
This paper provides a new method for the classification of rainfall areas in convective and stratiform rain using MSG/SEVIRI (Spinning Enhanced Visible and Infrared) data. The proposed approach is based on spectral and temporal properties of clouds. The spectral parameters used are: brightness temperature (BT) and brightness temperature differences (BTDs), and the temporal parameter (RCT10.8) is the rate of change of (BT) in the 10.8µm channel over two consecutive images. The developed rain area classification technique (RACT-DN) is based on two multilayer perceptron neural networks (MLP-D for daytime and MLP-N for nighttime) which relies on the correlation of satellite data with convective and stratiform rain. The two algorithms (MLP-D and MLP-N) are trained using as reference data from ground meteorological radar over northern Algeria. The results show that RACT-DN classifier gives accurate discrimination between convective and stratiform areas during daytime and nighttime.
Бесплатно
Image Watermarking in Frequency Domain using Hu's Invariant Moments and Firefly Algorithm
Статья научная
Preventing the digital content from being copied, manipulated and illegal ownership claims is one of the biggest challenges that appeared with the widespread usage of computing facilities. Watermarking is one way to tag a digital document with a watermark, perceptible or imperceptible, so as to later prove the ownership or authenticity of the document, in case the need arises. Robust and Fragile watermarking is used in case of proving ownership and authenticity, respectively. This paper proposes a watermarking approach based on Discrete Wavelet Transform (DWT), Hessenberg Decomposition (HD) and Singular Value Decomposition (SVD) approach, augmented with Firefly Algorithm (FA). To make the approach blind, the proposed technique uses Hu’s invariant moments which are invariant against rotation, scaling and translation (RST) attack over the image. In the resulting watermarked image, the watermark is imperceptible, which make it suitable for a large class of watermarking applications. In the proposed approach, a given colour image is subjected to 2 Level DWT for decomposing into sub-bands, namely LL, LH, HL and HH bands. These coefficients of HH band are fed as input for HD. The output is operated for SVD for obtain U, S and V matrices. The Hu’s invariant moments are scaled and mapped to binary string using logarithm scaling. The binary matrix, corresponding to binary watermark, is XoRed with the invariant moments, in a repeated manner, to obtain a new binary matrix, of the same dimension as count of 2X2 partitions of S. The watermark is embedded by changing the orthogonal V matrices. The magnitude of the change is computed with Firefly algorithm considering the robustness and imperceptibility as the trade-off parameters. The firefly algorithm is one of the nature inspired optimization algorithm. The proposed watermarking approach is capable of withstanding JPEG compression attack, filtering attacks and noise. PSNR and SSIM are used as the quality metric for accessing the watermarked image quality. It turns out that the proposed watermarking technique gives a considerable improvement over robustness and imperceptibility as compared to the benchmark approaches. The performance of the proposed approach as compared to the benchmark approach, increases in linear manner with the dimension of the image under consideration, reaching from 1 percent to 4 percent for image dimensions ranging from 400X400 to 1200X1200 pixels.
Бесплатно
Image retrieval based local motif patterns code
Статья научная
We present a new technique for content based image retrieval by deriving a Local motif pattern (LMP) code co-occurrence matrix (LMP-CM). This paper divides the image into 2 x 2 grids. On each 2 x 2 grid two different Peano scan motif (PSM) indexes are derived, one is initiated from top left most pixel and the other is initiated from bottom right most pixel. From these two different PSM indexes, this paper derived a unique LMP code for each 2 x 2 grid, ranges from 0 to 35. Each PSM minimizes the local gradient while traversing the 2 x 2 grid. A co-occurrence matrix is derived on LMP code and Grey level co-occurrence features are derived for efficient image retrieval. This paper is an extension of our previous MMCM approach [54]. Experimental results on popular databases reveal an improvement in retrieval rate than existing methods.
Бесплатно
Image semantic segmentation using deep learning
Статья научная
In the fields of Computer Vision, Image Semantic Segmentation is one of the most focused research areas. These are widely used for several real-time problems for finding the foreground or background scenes of a given image or a video. Initially, it is achieved using computer vision techniques, later once the deep learning is in its rise, ultimately it took over the entire image classification and segmentation techniques. These are widely surveyed and reviewed as they are used in several Image Processing, Feature Detection and Medical Fields. All the models for implementing Image Segmentation are mostly done using a specific neural network architecture called a convolution neural network. In this work, firstly we'll study the implementation of Image Segmentation models and advantages, disadvantages over one another including their development trends. We'll be discussing all the models and their applications concerning other fancy methods that are mostly used which involves hyperparameters and the transitive comparison between them.
Бесплатно
Статья научная
We are proposing a novel algorithm for tracking human face(s) in different background video sequences. We have trained both face and non-face images which help in face(s) detection process. At first, FAST features and corner points are extracted from the detected face(s). Further, mid points are calculated from corner points. FAST features, corner points and mid points are combined together. Using the combined points, point tracker tracks face(s) in the frames of the video sequence. Standard metrics were adopted for measuring the performance of the proposed algorithm. Low resolution video sequences with challenges such as partial occlusion, changes in expression, variations in illumination and pose took part while testing the proposed algorithm. Test results clearly indicate the robustness of the proposed algorithm on all different background challenging video sequences.
Бесплатно
Imaging Techniques for Cancer Diagnosis and Scope for Enhancement
Статья научная
Imaging techniques are used to create images of structure, function and pathology of human body organs for cancer diagnosis. Various imaging techniques like X-ray, Ultra-Sonography (US), Positron Emission Tomography (PET), Ultrasound, MRI etc. are used for cancer diagnosis. These imaging techniques have gone through Lot of advancements during lost few years. These techniques vary in the technology and application. Various artifacts exist in these imaging techniques and images produced by theses imaging techniques. These artifacts can be exploited to enhance the imaging technique and the images produced by them.
Бесплатно
Implementation of Hand Sign Recognition for Non-Linear Dimensionality Reduction based on PCA
Статья научная
Hand Sign or gesture recognition is the way of communication for hearing and speech impaired people. Gestures are formed from motion of body or state but commonly initiate from the hand and face. Speech and gestures are the expressions; these are the communication medium between human beings. Hand gesture is movement or motion of human hand. Gesture recognition is mathematical interpretation of human hand by using computing devices. There are different sign languages used in all over world and have its own grammar structure. Even in India has different languages used in every state, sign languages has little difference in contra dictionary region. Hand sign recognition is used for robot control applications and sign language interpretation.
Бесплатно
Статья научная
Future high resolution instrument planned by ISRO for space remote sensing will lead to higher data rates because of increase in resolution and dynamic range. Hence, image compression implementation becomes mandatory. Presently designed compression technique does not take account of imaging system noise like photon noise etc. This ignorance affects compression system performance. As a solution, this paper proposes MLG (Multi Linear Gain) operation prior to main compression system. With digital MLG operation, captured image can be optimally adjusted to systems noise. Proposed MLG operation improves compression ratio. Simulation results show 15-30% improvement in lossless compression ratio. However this improvement depends on MLG gains and corner points which can be driven by system SNR plot. MLG operation also helps in improving SNR at lower radiance input, when lossy JPEG2000 compression is used as main compression. Up to 1-6 dB SNR improvement is observed in simulations. Proposed MLG implementation is of very low complexity and planned to be used in future missions.
Бесплатно
Статья научная
This paper presents, complete step by step description design and implementation of a high speed technique for character segmentation of license plate based on thresholding algorithm. Because of vertical edges in the plate, fast Sobel edge detection has been used for extracting location of license plate, after stage edge detection the image is segmented by thresholding algorithm and the color of characters is changed to white and the color of background is black. Then, boundary’s pixels of license plate are scanned and their color is changed to black pixels. Afterward the image is scanned vertically and if the number of black pixels in a column is equal to the width of plate or a little few, then the pixels of that column is changed to white pixel, until create white columns between characters, in continue we change pixels around license plate to white pixels. Finally characters are segmented cleanly. We test proposed character segmentation algorithm for stage recognition of number by code that we design. Results of experimentation on different images demonstrate ability of proposed algorithm. The accuracy of proposed character segmentation is 99% and average time of character segmentation is 15ms with thresholding algorithm code and 0.7ms only segmentation character code that is very small in comparison with other algorithms.
Бесплатно
Implementing Blind De-convolution with Weights on X-ray Images for Lesser Ringing Effect
Статья научная
X-rays and other medical images are distorted because of the limitations of the Imaging system. The other source from where the distortions get in are the transmission channels. The distortions are generally noise and blur. Unless and until the medical images are free of noise and blur they cannot be used by medical professionals to the full extent for diagnosis purpose. Therefore these images must be restored properly before they are used for diagnosis purpose. There are different restoration techniques out of which one is Blind Image Deconvolution. X-ray images restored with this technique have ringing effect in them. Using edgetaper (matlab function) prior to Blind Image Deconvolution reduces the ringing effect to an extent. This paper presents Blind Deconvolution algorithm with weights which gives lesser ringing effect in X-ray images when they are restored.
Бесплатно
Implementing of microscopic images mosaic revising algorithm
Статья научная
Microscopic image mosaic stitches several adjacent images into an integrated seamless picture, and is of significant practical value to remote medicine applications, especially in remote diagnosis. However, due to limitation in image acquisition method, a mismatch could occur as a result of variance in adjacent image stitching data and accumulation of errors. The current image stitching method still has room for improvement regarding processing speed and effectiveness, particularly in precision. In this paper, we proposed a new image mosaic revising algorithms based on the relativity of adjacent images and expounding the principal and equations on image mosaic error revising, as well as achieving automatic intelligent calculation with the revised algorithm. Through experiment, inaccurate pathological mosaic images from 20 groups were revised rapidly and accurately with error controlled within one pixel. It was proved that the approach is effective in revising the error matching in microscopic images mosaic. Moreover, it is easy to operate and effective for more accurate image stitching.
Бесплатно
Статья научная
In this paper, we propose two algorithms based on the popular Bit Plane Splicing Least Significant Bit (LSB) Technique for secret data hiding inside images. One major disadvantage of Bit Plane Splicing LSB technique is its low hiding capacity which results in severe degradation of the cover image upon hiding large amount of data. The proposed algorithms overcome this issue by imposing hiding rules based on the intensity level of pixels. In addition the method for data hiding is done in a non sequential manner using linear congruent random number generators. Experiment results show that the proposed techniques called Optimum Intensity Based Distributed Hiding (OIBDH) technique and Linear Congruent Optimum Intensity Based Distributed Hiding with Key (LC-OIBDH-k) outperforms Bit Plane Splicing LSB technique as they have better hiding capacity with less degradation in the cover image. Furthermore, the proposed algorithms are tested using absolute entropy curves and results show that our proposed techniques have lower absolute entropy difference compared to Bit Plane Splicing LSB technique in all the tested images and for different secret data sizes.
Бесплатно
Improved Cryptanalysis of CMC Chaotic Image Encryption Scheme
Статья научная
Recently, chaos has attracted much attention in the field of cryptography. To study the security with a known image of a symmetric image encryption scheme, the attack algorithm of equivalent key is given. We give the known image attacks under different other conditions to obtain the equivalent key. The concrete step and complexity of the attack algorithm is given. So the symmetric image encryption scheme based on 3D chaotic cat maps is not secure.
Бесплатно
Статья научная
The purpose of speech emotion recognition system is to classify speaker's utterances into different emotional states such as disgust, boredom, sadness, neutral and happiness. Speech features that are commonly used in speech emotion recognition (SER) rely on global utterance level prosodic features. In our work, we evaluate the impact of frame-level feature extraction. The speech samples are from Berlin emotional database and the features extracted from these utterances are energy, different variant of mel frequency cepstrum coefficients (MFCC), velocity and acceleration features. The idea is to explore the successful approach in the literature of speaker recognition GMM-UBM to handle with emotion identification tasks. In addition, we propose a classification scheme for the labeling of emotions on a continuous dimensional-based approach.
Бесплатно
Improved Parallel Lane Detection Using Modified Additive Hough Transform
Статья научная
Lane detection has recognition in real time vehicular ad-hoc system. That study work concentrate on giving greater efficiency in lane detection by utilizing the additive Hough transform to identify the curve lanes and convert into data parallelism in order to improve the speed of the proposed technique by using fork and join process. To accomplish performance evaluation various metrics is likely to be considered. The performance of lane detection algorithms is generally evaluated in terms of algorithm results and parallel results. Algorithm results is evaluated in terms of accuracy, error rate, execution time ,overhead and parallel results is evaluated in terms of speed, efficiency etc. To complete performance comparison the result of proposed algorithm is going to be compared with existing lane detection algorithms. Intelligent transportation systems are available these days for increasing the safety of the vehicles and reduce incident ratio. A new technique which uses modified additive hough transform is used to reduce the limitations of existing technique. The proposed algorithm has been designed and implemented in MATLAB.
Бесплатно
