Narrow Results

Image compression is the emerging field to transmit the multimedia products like image, video and audio. Image compression is used to reduce the storage quantity as much as possible. The objective of this paper is to compare the multiscale transform based image compression encoding techniques. The multiscale transforms involved in this paper are wavelet transform, bandelet transform, curvelet transform, ridgelet transform and contourlet transform. Wavelet transform allows good localization both in time and frequency domain. Bandelet transform takes geometric regularity of the natural images to improve the efficiency of representation. Curvelet transform handles curve discontinuities well. Curvelets are the good tool for the analysis and the computation of partial differential equations. Curvelets also have micro local features which make them especially adapted to certain reconstruction problems with missing data. The ridgelet transform is the core idea behind curvelet transform. It is used to represent objects with line singularities. The contourlet transform gets smooth contours and edges at any orientation. It filters noise as well. The Encoding techniques involved in this paper are spatial orientation tree wavelet (STW), set partitioned embedded block (SPECK) and compression with reversible embedded wavelet (CREW). The performance parameters such as peak signal to noise ratio (PSNR), image quality index and structural similarity index (SSIM) are used for the purpose of evaluation. It is found that bandelet transform with all the encoding techniques work well.

Image denoising, a significant research area in the field of medical image processing, makes an effort to recover the original image from its noise corrupted image. The Pulse Coupled Neural Networks (PCNN) works well against denoising a noisy image. Generally, image denoising techniques are directly applied on the pixels. From the literature review, it is reported that denoising after frequency domain transformation is performing better since noise removal is applied over the coefficients. Motivated by this, in this paper, a new technique called the Static Thresholded Pulse Coupled Neural Network (ST-PCNN) is proposed by combining PCNN with traditional filtering or threshold shrinkage technique in Contourlet Transform domain. Four different existing PCNN architectures, such as Neuromime Structure, Intersecting Cortical Model, Unit-Linking Model and Multichannel Model are considered for comparative analysis. The filters such as Wiener, Median, Average, Gaussian and threshold shrinkage techniques such as Sure Shrink, HeurShrink, Neigh Shrink, BayesShrink are used. For noise removal, a mixture of Speckle and Gaussian noise is considered for a CT skull image. A mixture of Rician and Gaussian noise is considered for MRI brain image. A mixture of Speckle and Salt and Pepper noise is considered for a Mammogram image. The Performance Metrics such as Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), Image Quality Index (IQI), Universal Image Quality Index (UQI), Image Enhancement Filter (IEF), Structural Content (SC), Correlation Coefficient (CC), and Weighted Signal-to-Noise Ratio (WSNR) and Visual Signal-to-Noise Ratio (VSNR) are used to evaluate the performance of denoising.

The fluid lens cameras are ideal candidates for medical imaging. Since no moving part is needed to change the focal length for focusing, the size of these generation of cameras are considerably smaller than the traditional glass lens cameras. The major drawback is that only one color channel can be in total focus, leaving the other two channels blurry, causing loss of edges and high frequency data of the two channels. In this paper, we use directional filters and contourlet transformation to retrieve the high frequency data to enhance the quality of the other two color channels. It is shown that it is important to choose the optimum number of contourlet decomposition level and also we will try to find a criteria to find the optimal number of levels. Using the proposed method on general and endoscopic images shows that the proposed algorithm yields to average PSNR improvement of 6.3db for endoscopic images which is 5.7db more than the previous proposed algorithm.

Coping with geometrical attacks in transform domain is crucial when we design a robust image watermarking scheme. In this paper, a novel contourlet-domain image watermarking scheme, which is robust to common signal processing and geometrical attacks, is proposed. First, the region with maximum energy in the directional subbands is considered for watermarking, i.e. the watermark can be embedded into the significant region as well as highly textured region of the host image. Then, for each coefficient of the selected subband, the strength factor was adaptively adjusted in terms of the energy of its parent and neighbor coefficients. Consequently, the tradeoff between the transparency and robustness of watermark can be achieved. Furthermore, the robust feature points, which can survive various signal processing and affine transformation, are extracted by using the Harris–Laplace detector. In watermark detection, the geometrical distortion of image is identified by using the feature template constructed during embedding phase, i.e. watermark resynchronization is performed. Experimental results show that the proposed watermark scheme is invisible and robust against common signal processing such as median filtering, sharpening, noise adding and JPEG compression, etc and geometrical attacks such as rotation, translation, scaling, row or column removal, shearing and local random bend, etc.

Motion estimation is a means, which consists in studying the displacement of objects in a video sequence, seeking the correlation between two successive frames, to predict the change in the contents position. Motion estimation is becoming a progressively significant requirement in a variety of applications such as medicine, robotics and video compression. In recent years, wavelets are effective tools for motion estimation, but the DWT (Discrete Wavelet Transform) will suffer from problems like translation sensitivity, poor directionality and absence of phase information. These three disadvantages make classical wavelets incapable of calculating motion in complex sequences (contain several directions.). In order to improve these negative aspects, we will choose geometric wavelet. Therefore, our objective is to propose a method capable of estimating the motion in terms of performance (speed and accuracy). This method will be based on the geometric wavelet transform and more precisely on the Contourlet transform. This work consists of two parts: in the first stage, the denoising process is examined by the Contourlet transform to ensure the precision of motion; in the second phase, we applied the iterative method of Horn and Schunck to calculate the motion in order to guarantee good speed. Comparative experimental results of artificial sequences show that the proposed algorithm obtains considerably better performance than several state-of-the-art methods.

Fingerprint verification systems have attracted much attention in secure organizations; however, conventional methods still suffer from unconvincing recognition rate for noisy fingerprint images. To design a robust verification system, in this paper, wavelet and contourlet transforms (CTS) were suggested as efficient feature extraction techniques to elicit a coverall set of descriptive features to characterize fingerprint images. Contourlet coefficients capture the smooth contours of fingerprints while wavelet coefficients reveal its rough details. Due to the high dimensionality of the elicited features, across group variance (AGV), greedy overall relevancy (GOR) and Davis–Bouldin fast feature reduction (DB-FFR) methods were adopted to remove the redundant features. These features were applied to three different classifiers including Boosting Direct Linear Discriminant Analysis (BDLDA), Support Vector Machine (SVM) and Modified Nearest Neighbor (MNN). The proposed method along with state-of-the-art methods were evaluated, over the FVC2004 dataset, in terms of genuine acceptance rate (GAR), false acceptance rate (FAR) and equal error rate (EER). The features selected by AGV were the most significant ones and provided 95.12% GAR. Applying the selected features, by the GOR method, to the modified nearest neighbor, resulted in average EER of $<1$%, which outperformed the compared methods. The comparative results imply the statistical superiority ($p<0.05$) of the proposed approach compared to the counterparts.

This paper presents the method of multi-resolution analysis used in 2D image data to extract the curved edge features. The method is based on the combination of multi-resolution decomposition through Wavelet Packet and Prime Ridgelet transform. We call this combination Prime Wavelet Packet Contourlet Transform-PWPC. At each leave of Packet Wavelet Packet Tree, the prime ridgelet transform is applied on the band pass image or packet, which contains the high frequency data. The experiment shows that the PWPC coefficients are good approximations to curved edges. The speed of PWPC is faster than that of the basic Curvelet transform. This transform is very suitable to represent the noisy curved features that often exist in medicine or nano/micro images.