Narrow Results

Vehicle type/make recognition based on images captured by surveillance cameras is a challenging task in intelligent transportation system and automatic surveillance. In this paper, we comparatively studied two feature extraction methods for image description, i.e. a new multiresolution analysis tool called Fast Discrete Curvelet Transform and the pyramid histogram of oriented gradients (PHOG). Curvelet Transform has better directional and edge representation abilities than widely used wavelet transform, which is particularly appropriate for the description of images rich with edges. PHOG represents the local shape by a histogram of edge orientations computed for each image sub-region, quantized into a number of bins, thus has the ascendency in its description of more discriminating information. A composite feature description from PHOG and Curvelet can further increase the accuracy of classification by taking their complementary information. We also investigated the applicability of the Rotation Forest (RF) ensemble method for vehicle classification based on the combined features. The RF ensemble contains a set of base multilayer perceptrons which are trained using principal component analysis to rotate the original axes of combined features of vehicle images. The class label is assigned by the ensemble via majority voting. Experimental results using more than 600 images from 21 makes of cars/vans show the effectiveness of the proposed approach. The composite feature is better than any single feature in the classification accuracy and the ensemble model produces better performance compared to any of the individual neural network base classifier. With a moderate ensemble size of 20, the Rotation Forest ensembles offers a classification rate close to 96.5%, exhibiting promising potentials for real-life applications.

Iris recognition is one of the most reliable personal identification methods. This paper presents a novel algorithm for iris recognition encompassing iris segmentation, fusion of statistical and co-occurrence features extracted from the curvelet and ridgelet transformed images. In this work, the pupil and iris boundaries are detected by using the equation of circle from three points on its circumference. Using Canny edge detection, the iris radius value is empirically chosen based on rigorous experimentation. Eyelash removal is done by using a horizontal 1-D rank filter. Iris normalization is done by mapping the detected iris region from the polar domain to the rectangular domain and the multi-resolution transforms such as curvelet and ridgelet transforms are applied for multi-resolutional feature extraction. The classification is done using Manhattan distance (M_d) and multiclass classifier with logistic function and the two results are compared. Here, the benchmark database CASIA-IRIS-V3 (Interval) is used for identification and recognition. It is observed that the ridgelet transform increases the iris recognition rate.

In this paper, we have proposed and presented a method for object tracking in video that exploits new tight frame of curvelets. We constructed a method of tracking that provides a sparse expansion for typical images having smooth contours. We used curvelet coefficients to segment and tracking of object in the sequence of frames. Results obtained for tracking of moving objects in video clips demonstrated an improved performance over other recent related methods available in the literature.

This paper proposes two different methods for fusing multispectral (MS) and panchromatic (PAN) satellite images using the nonsubsampled contourlet transform (NSCT). The NSCT decomposes the images into several directional subbands at flexible resolutions. The finer subbands of the MS image decomposition, which represent the high frequency details, are modified with the subbands from PAN image. The inverse transform gives a high resolution MS representation of the given low resolution MS image. The paper also proposes the use of the parameters mean structural similarity index measure (MSSIM) and edge stability mean square error (ESMSE) to measure the quality of the fused image, in addition to standard parameters like correlation coefficient, PSNR, RASE and ERGAS. Experiments show that the proposed methods outperform standard fusion techniques in terms of both visual quality and quantitative error measures. The fused images obtained using the proposed methods show good structural similarity and better edge stability as the contourlet transform very well extracts the oriented edge details from the high resolution PAN image.

Abdomen related diseases are responsible of many deaths every year. These deaths can be reduced by early diagnosis of abdomen diseases. Computer aided diagnosis (CAD) can play vital role in early detection of diseases. Hence, a novel CAD is proposed in this paper that can diagnose abdomen diseases like Hepatocellular carcinoma, cysts and Calculi using statistical curvelet texture descriptors. The proposed CAD is divided into four stages: (a) Image segmentation using active contours, (b) feature extraction, (c) feature selection and (d) abdomen disease classification. The regions of interest (ROIs) are segmented from 120CT images using active contour models. The statistical features are extracted from segmented ROIs. Further, the classifiers are used to evaluate the ability of feature set in diagnosis various diseases of abdomen. The performance metrics indicates that the proposed CAD achieves accuracy of 87.9% using curvelet coefficient features and neural network as classifier.

Texture classification has long been an important research topic in image processing. Nowadays classification based on wavelet transform is being very popular. Wavelets are very effective in representing objects with isolated point singularities, but failed to represent line singularities. Recently, ridgelet transform which deal effectively with line singularities in 2D is introduced. But images often contain curves rather than straight lines, so curvelet transform is designed to handle it. It allows representing edges and other singularities along lines in a more efficient way when compared with other transforms. In this paper, the issue of texture classification based on curvelet transform has been analyzed. Features are derived from the sub-bands of the curvelet decomposition and are used for classification for the four different datasets containing 20, 30, 112 and 129 texture images respectively. Experimental results show that this approach allows high degree of success rate in classification to be obtained.

In many practical situations, magnetic resonance imaging (MRI) needs reconstruction of images at low measurements, far below the Nyquist rate, as sensing process may be very costly and slow enough so that one can measure the coefficients only a few times. Segmentation of such subsampled reconstructed MR images for medical analysis and diagnosis becomes a challenging task due to the inherent complex characteristics of the MR images. This paper considers reconstruction of MR images at compressive sampling (or compressed sensing (CS)) paradigm followed by its segmentation in an integrated platform. Image reconstruction is done from incomplete measurement space with random noise injection iteratively. A weighted linear prediction is done for the unobserved space followed by spatial domain denoising through adaptive recursive filtering. The reconstructed images, however, suffer from imprecise and/or missing edges, boundaries, lines, curvatures etc. and residual noise. Curvelet transform (CT) is purposely used for removal of noise and for edge enhancement through hard thresholding and suppression of approximate subbands, respectively. Then a fuzzy entropy-based clustering, using genetic algorithms (GAs), is done for segmentation of sharpen MR Image. Extensive simulation results are shown to highlight performance improvement of both image reconstruction and segmentation of the reconstructed images along with relative gain over the existing works.

Due to the huge advancement in technology, digitizing the multimedia content like text, images and videos has become easier. Everyday huge amounts of multimedia content are shared through the social networks using internet. Sometimes this multimedia content can be hacked by the hackers. This will lead to the misuse of the data. On the other hand, the medical content needs high security and privacy. Motivated by this, joint secured medical image compression–encryption mechanisms are proposed in this paper using multiscale transforms and symmetric key encryption techniques. The multiscale transforms involved in this paper are wavelet transform, bandelet transform and curvelet transform. The encryption techniques involved in this paper are international data encryption algorithm (IDEA), Rivest Cipher (RC5) and Blowfish. The encoding technique used in this paper is embedded block coding with truncation (EBCOT). Experimental results are done for the proposed works and evaluated by using various parameters like Peak Signal-to-Noise Ratio (PSNR), Mean Square Error (MSE), Image Quality Index (IQI) and Structural Similarity Index (SSIM), Average Difference (AD), Normalized Cross-correlation (NK), Structural Content (SC), Maximum difference (MD), Laplacian Mean Squared Error (LMSE) and Normalized Absolute Error (NAE). It is justified that the proposed approaches in this paper yield good results.

Images captured under low light are noisy and consist of unidentifiable features. Low light noise problem occurs in imaging devices because of smaller sensor size or insufficient exposure. Low light image denoising is an exacting task in many image processing applications. This paper proposes a patch-based image denoising method for low light images in the curvelet domain with contrast enhancement. Curvelet transform is a directional transform and it gives the best sparse representation for images with edges. Here the Expectation–Maximization (EM) algorithm, based on the Gaussian mixture adaptation method is performed in the curvelet domain for denoising. EM Algorithm helps in computing the Gaussian mixture model (GMM) parameters from the patches which are used in maximum a posteriori estimation to update them. GMM parameters and patches are updated periodically until a satisfactory result is achieved. Simulation is performed on standard test data set, and then extended to natural low light noisy images. The results of the proposed technique are then compared using quality metrics such as Peak Signal to Noise Ratio and Structural Similarity Index. It is observed that the use of curvelet transform in denoising process helps to restore the structural information satisfactorily.

In this paper, we have introduced a novel integral transform namely linear canonical curvelet transform (LCCT). Firstly, we established basic properties of LCCT including admissibility condition, Moyals principle, inversion formula and range theorem. Toward the culmination of the paper, we formulate a couple of Heisenberg-type inequalities associated with LCCT.

The curvelet transform of a tempered distribution is defined as an infinitely differentiable function of (a, b, θ) with a polynomial growth in b. An inversion formula of the curvelet transform on tempered distributions is also obtained.

In recent years, multi-scale transform application in image processing especially for magnetic resonance (MR) images has been raised. Wavelet transform is introduced as a useful tool in image processing and it is capable of effectively removing noise from magnetic resonance images. The main problem with wavelet transform is that it is not able to distinguish one dimensional (1D) or higher dimentional discontinuities in images. In other words, since the wavelet transform is two dimensional (2D), it is considered as a separable transform, it is solely able to identify pointwise discontinuity in images. A proposed solution for this issue is an inseparable transform which is named curvelet. Time frequency transform based noise elimination methods, usually rely on thresholding. There are two important factors involved in thresholding: (1) a method to determine the threshold limit, (2) the implementation of the threshold. In curvelet method, by setting a hard threshold at low levels of noise the obtained similarity index is 0.9254. The proposed method for noise elimination and edge detection in this paper is applying curvelet transform in combination with wavelet transform, which on average leads to 10% improvement compared with wavelet method. The results show the efficiency of this method in different parts of image processing on simulated and actual MR images.

A novel feature extraction method with global-locally preserving projections based on multi-scale transform (MGLPP) is proposed. The proposed method consists of the following steps. First, the original images are decomposed using curvelet transform into several elements. Second, the proposed GLPP algorithm is applied to the curvelet coefficients of detailed levels to determine the optimal projection vector W. Finally, the distance between the training samples and testing samples in low-dimensional space is calculated to recognize and classify the target. Moreover, the introduction of degree of belonging functions of the fuzzy c-means (FCM) method to the proposed MGLPP method aims to identify the k neighbors of each sample. The experimental results demonstrate that the MGLPP method performs better in dimensionality reduction and recognition accuracy than the traditional principal component analysis (PCA) and local preserving projection algorithms (LPP).