8305233223

High false positive face detection is a crucial problem which leads to low performance face recognition in surveillance system. The performance can be increased by reducing these false positives so that non-face can be discarded first prior to recognition. This paper presents a combination of two well-known algorithms, Adaboost and Neural Network, to detect face in static images which is able to reduce the false-positives drastically. This method utilizes Haar-like features to extract the face rapidly using integral image. A cascade Adaboost classifier is used to increase the face detection speed. Due to using only this cascade Adaboost produces high false-positives, neural network is used as the final classifier to verify face or non-face. For a faster processing time, hierarchical Neural Network is used to increase the face detection rate. Experiments on four different face databases, which consist more than one thousand images, have been conducted. Results reveal that the proposed method achieves about 93.34% of detection rate and 0.34% of false positives compared to original cascade Adaboost method which achieves about 98.13% of detection rate with 6.50% of false positives. The processed images size is 240 x 320 pixels. Each frame is processed at about 2.25 sec which is slights lower than the original method, which only takes about 0.82 sec.

Biometric technology plays a vital role for providing the security which is imperative part in secure system. Human face recognition is a potential method of biometric authentication. This paper presents a process of face recognition system using principle component analysis with Back propagation neural network where features of face image has been combined by applying face detection and edge detection technique. In this system, the performance has been analyzed based on the proposed feature fusion technique. At first, the fussed feature has been extracted and the dimension of the feature vector has been reduced using Principal Component Analysis method. The reduced vector has been classified by Back propagation neural network based classifier. In recognition stage, several steps are required. Finally, we analyzed the performance of the system for different size of the train database. The performance analysis shows that the efficiency has been enhanced when the feature extraction operation performed successfully. The performance of the system has been reached more than 92% for the adverse conditions.

Facial feature detection and extraction plays a significant role in face recognition. In this paper, we proposed a new approach for face feature description using support vector machine classifier. Searching image from the image contents is a major challenge in computer vision and pattern recognition. The textual label is assigned to the face feature according to the semantic meaning of the feature. The concept of the textual image description gives advantages to the user to search the image with the help of the visual attributes of the image. The face features like eyes, nose, and mouth are labeled with the visual attributes like small, normal and large.

Signatures are imperative biometric attributes of humans that have long been used for authorization purposes. Most organizations primarily focus on the visual appearance of the signature for verification purposes. Many documents, such as forms, contracts, bank cheques, and credit card transactions require the signing of a signature. Therefore, it is of upmost importance to be able to recognize signatures accurately, effortlessly, and in a timely manner. In this work, an artificial neural network based on the well-known Back-propagation algorithm is used for recognition and verification. To test the performance of the system, the False Reject Rate, the False Accept Rate, and the Equal Error Rate (EER) are calculated. The system was tested with 400 test signature samples, which include genuine and forged signatures of twenty individuals. The aim of this work is to limit the computer singularity in deciding whether the signature is forged or not, and to allow the signature verification personnel to participate in the deciding process through adding a label which indicates the amount of similarity between the signature which we want to recognize and the original signature. This approach allows judging the signature accuracy, and achieving more effective results.

Content Based Image Retrieval (CBIR) is a technique that enables a user to extract an image based on a query, from a database containing a large amount of images. A very fundamental issue in designing a content based image retrieval system is to select the image features that best represent the image contents in a database. In this paper, our proposed method mainly concentrated on database classification and efficient image representation. We present a method for content based image retrieval based on support vector machine classifier. In this method the feature extraction was done based on the colour string coding and string comparison. We succeed in transferring the images retrieval problem to strings comparison. Thus the computational complexity is decreases obviously. The image database used in our experiment contains 1800 colour images from Corel photo galleries.This CBIR approach has significantly increased the accuracy in obtaining results for image retrieval.

The effective content-based image retrieval (CBIR) needs efficient extraction of low level features like color, texture and shapes for indexing and fast query image matching with indexed images for the retrieval of similar images. Features are extracted from images in pixel and compressed domains. However, now most of the existing images are in compressed formats like JPEG using DCT (discrete cosine transformation). In this paper we study the issues of efficient extraction of features and the effective matching of images in the compressed domain. In our method the quantized histogram statistical texture features are extracted from the DCT blocks of the image using the significant energy of the DC and the first three AC coefficients of the blocks. For the effective matching of the image with images, various distance metrics are used to measure similarities using texture features. The analysis of the effective CBIR is performed on the basis of various distance metrics in different number of quantization bins. The proposed method is tested by using Corel image database and the experimental results show that our method has robust image retrieval for various distance metrics with different histogram quantization in a compressed domain.

Content based image retrieval technique is done by three primitive methods namely through color, shape and texture. This paper provides specified path to use these primitive features to retrieve the desired image. The technique by which we obtain the required image is CBIR. In CBIR first the HSV color space is quantified to obtain the color histogram and texture features. Using these components a feature matrix is formed. Then this matrix is mapped with the characteristic of global color histogram and local color histogram, which are analysed and compared. For the cooccurrence matrix between the local image and the images in the database to retrieve the image. For extracting shape feature gradient method is used here. Based on this principle, CBIR system uses color, texture and shape fused features to retrieve desired image from the large database and hence provides more efficiency or enhancement in image retrieval than the single feature retrieval system which means better image retrieval results.

The image descriptors based on multi-features fusion have better performance than that based on simple feature in content-based image retrieval (CBIR). However, these methods still have some limitations: 1) the methods that define directly texture in color space put more emphasis on color than texture feature; 2) traditional descriptors based on histogram statistics disregard the spatial correlation between structure elements; 3) the descriptors based on structure element correlation (SEC) disregard the occurring probability of structure elements. To solve these problems, we propose a novel image descriptor, called Global Correlation Descriptor (GCD), to extract color and texture feature respectively so that these features have the same effect in CBIR. In addition, we propose Global Correlation Vector (GCV) and Directional Global Correlation Vector (DGCV) which can integrate the advantages of histogram statistics and SEC to characterize color and texture features respectively. Experimental results demonstrate that GCD is more robust and discriminative than other image descriptors in CBIR

A novel content-based image retrieval (CBIR) schema with wavelet and color features followed by ant colony optimization (ACO) feature selection has been proposed in this paper. A new feature extraction schema including texture features from wavelet transformation and color features in RGB and HSV domain is proposed as representative feature vector for images in database. Also, appropriate similarity measure for each feature is presented. Retrieving results are so sensitive to image features used in content-based image retrieval. We address this problem with selection of most relevant features among complete feature set by ant colony optimization based feature selection. To evaluate the performance of our proposed CBIR schema, it has been compared with older proposed systems, results show that the precision and recall of our proposed schema are higher than older ones for the majority of image categories

In this letter, we propose a new adaptive weighted mean filter (AWMF) for detecting and removing high level of salt-and-pepper noise. For each pixel, we firstly determine the adaptive window size by continuously enlarging the window size until the maximum and minimum values of two successive windows are equal respectively. Then the current pixel is regarded as noise candidate if it is equal to the maximum or minimum values, otherwise, it is regarded as noise-free pixel. Finally, the noise candidate is replaced by the weighted mean of the current window, while the noise-free pixel is left unchanged. Experiments and comparisons demonstrate that our proposed filter has very low detection error rate and high restoration quality especially for high-level noise.

Image Enhancement is a process of improving the quality of an image by improving the contrast of the images. Images acquired underwater usually suffers from non-uniform illumination, low-contrast, motion blur effect due to turbulence in the flow of water, in under ocean environment, scattering of light from different particles of various sizes, diminished intensity and color level due to poor visibility conditions, suspended moving particles and so on. Due to all these factors, various types of noises occur and to reduce the effects arising out of these factors, a number of methods are required to be incorporated to improve the quality of underwater images. The present paper shows a comparative study of the various image enhancement techniques used for enhancing underwater images and introduces a suitable novel hybrid method for improving the image quality.

A novel edge-preserving image decomposition based on saliency map is proposed to avoid halo phenomena in traditional image decompositions. Our method uses feature of saliency map that emphasizes salient edges in image to determine whether a region should be retained or processed, and then blur the image adaptively according to the local saliency average gradient. Experiments show that the proposed decomposition performs well in smoothing and enhancement. We compare our results with the guided filter and bilateral filter, and demonstrate a variety of applications including HDR and multi-scale enhancement.

Steganography is a data hiding technique that is widely used in various information securing applications. Steganography transmits data by hiding the existence of the message so that a viewer cannot identify the transmission of message and hence not able to decrypt it. This work proposes a data securing technique that is used for hiding multiple color images into a single color image using the Discrete Wavelet Transform. The cover image is split up into R, G and B planes. Secret images are embedded into these planes. An N-level decomposition of the cover image and the secret images are done and some frequency components of the same are combined. Secret images are then extracted from the stego image. Here, the stego image obtained has a less perceptible changes compared to the original image with high overall security.

In this paper, we introduce a ghost-free High Dynamic Range imaging algorithm for obtaining ghost-free high dynamic range (HDR) images. The multiple image fusion based HDR method work only on condition that there is no movement of camera and object when capturing multiple, differently exposed low dynamic range (LDR) images. The proposed algorithm makes three LDR images from a single input image to remove such an unrealistic condition. For this purpose a histogram separation method is proposed in the algorithm for generating three LDR images by stretching each separated histogram. An edge-preserving denoising technique is also proposed in the algorithm to suppress the noise that is amplified in the stretching process. In the proposed algorithm final HDR image free from ghost artifacts in dynamic environment because it self-generates three LDR images from a single input image. Therefore, the proposed algorithm can be use in mobile phone camera and a consumer compact camera to provide the ghost artifacts free HDR images in the form of either inbuilt or post-processing software application.

This paper presents an efficient algorithm for solving a balanced regularization problem in the frame-based image restoration. The balanced regularization is usually formulated as a minimization problem, involving an 2 data-fidelity term, an 1 regularizer on sparsity of frame coefficients, and a penalty on distance of sparse frame coefficients to the range of the frame operator. In image restoration, the balanced regularization approach bridges the synthesis-based and analysis-based approaches, and balances the fidelity, sparsity, and smoothness of the solution. Our proposed algorithm for solving the balanced optimal problem is based on a variable splitting strategy and the classical alternating direction method. This paper shows that the proposed algorithm is fast and efficient in solving the standard image restoration with balanced regularization. More precisely, a regularized version of the Hessian matrix of the 2 data-fidelity term is involved, and by exploiting the related fast tight Parseval frame and the special structures of the observation matrices, the regularized Hessian matrix can perform quite efficiently for the frame-based standard image restoration applications, such as circular deconvolution in image deblurring and missing samples in image inpainting. Numerical simulations illustrate the efficiency of our proposed algorithm in the frame-based image restoration with balanced regularization

Image deblurring is one of the fundamental problems in the image processing and computer vision fields. In this paper, we propose a new approach for restoring images corrupted by blur and impulse noise. The existing methods used to address this problem are based on minimizing the objective functional, which is the sum of the L1-data fidelity term, and the total variation (TV) regularization term. owever, TV introduces staircase effects. Thus, we propose a new objective functional that combines the tight framelet and TV to restore images corrupted by blur and impulsive noise while mitigating staircase effects. The minimization of the new objective functional presents a computational challenge. We propose a fast minimization algorithm by employing the augmented Lagrangian technique. The experiments on a set of image deblurring benchmark problems show that the proposed method outperforms previous state-of-the-art methods for image restoration.

In this paper, a novel and simple restorationbased fog removal approach is proposed. Here, we proposed an approach which is based on gamma transformation method and median filter. Transmission map is refined by the gamma transformation method. Then obtained transmission map is smoothed by a median filter. Qualitative and quantitative analysis demonstrate that proposed algorithm performs well in comparison with bilateral filtering. It can handle color as well as gray images. Proposed algorithm, due to its speed and ability to improve visibility, may be used in many systems, ranging from tracking and navigation, surveillance, consumer electronics, intelligent vehicles to remote sensing.

This paper deals with estimation of parameters for motion blurred images. The objectives are to estimate the length (L) and the blur angle () of the given degraded image as accurately as possible so that the restoration performance can be optimised. Gabor filter is utilized to estimate the blur angle whereas a trained radial basis function neural network (RBFNN) estimates the blur length. Once these parameters are estimated the conventional restoration is performed. To validate the proposed scheme, simulation has been carried out on standard images as well as in real images subjected to different blur angles and lengths. The robustness of the scheme is also validated in noise situations of different strengths. In all situations, the results have been compared with standard schemes. It is in general observed that the proposed scheme outperforms its counterparts in terms of restoration parameters and visual quality.

The restoration of motion blurred images is a hot shot in the field of image processing. In this paper, The degradation model of motion blurred images is clarified, and the estimation of the PSF parameter as well as its algorithm is presented. Frequency spectrum and Radon transition are used for the length and angle calculation, then Wiener filtering and Inverse filtering are used for recovering the noisy and motion blurred images. The matrix conjugate gradient is used to process the images which are processed by Wiener filtering and Inverse filtering. These simulation results indicate that matrix conjugate gradient is effective and promising for their recovery.

A new model for impulse noise reduction with a recursive and noise-exclusive fuzzy switching median filter is proposed. The filter uses a S-type membership function for fuzzifying the noisy input variable and then estimates a correction label for it that aims at canceling the noisy effect. The fuzzification process provides better smoothing and generalization that improve the performance of the filter. The recursive and noise-exclusive operations further enhance the noise reduction capability of the filter. The recursive operation replaces the current pixel with the filtered pixel(s) and the noise-exclusive filtering uses only noise-free pixel(s) of the working window. The net effect of the filtering process thus preserves the sharp edges and fine details of the image in a more effective manner. The superiority of the proposed model to others in removing high density noise is established both quantitatively and qualitatively with various benchmarks and real-time test data sets. With the incorporation of the functionalities like fuzzy reasoning and noise-exclusive operations in the filtering process, the model becomes more efficient and fast. The applicability of the model thus opens ample scopes for hardware implementation in many electronics products that deal with images and videos.

Satellite imaging is being the most attractive source of information for the governmental agencies and the commercial companies in last decade. The quality of the images is very important especially for the military or the police forces to pick the valuable information from the details. Satellite images may have unwanted signals called as noise in addition to useful information for several reasons such as heat generated electrons, bad sensor, wrong ISO settings, vibration and clouds. There are several image enhancement algorithms to reduce the effects of noise over the image to see the details and gather meaningful information. Many of these algorithms accept several parameters from the user to reach the best results. In the process of denoising, there is always a competition between the noise reduction and the fine preservation. If there is a competition between the objectives then an evolutionary multi objective optimization (EMO) is needed. In this work, the parameters of the image denoising algorithms have been optimized to minimize the trade-off by using improved Strength Pareto Evolutionary Algorithm (SPEA2). SPEA2 differs from the other EMO algorithms with the fitness assignment, the density estimation and the archive truncation processes. There is no single optimal solution in a multi objective problems instead there is a set of solutions called as Pareto efficient. Four objective functions, namely Mean Square Error (MSE), Entropy, Structural SIMilarity (SSIM) and Second Derivative of the image, have been used in this work. MSE is calculated by taking the square of difference between the noise free image and the deniosed image. Entropy is a measure of randomness of the content of difference image. The lower entropy is the better. The second derivate of an image can be achieved by convolving the image with the Laplacian Mask. SSIM algorithm is based on the similarities of the structures on the noise free image and the structures of the denoised image. For the image enhancement algorithms, Insight Segmentation and Registration Toolkit (ITK) is selected. ITK is an open source project and it is being developed in C++ to provide developers with a rich set of applications for image analysis. It includes tens of image filters for the registration and segmentation purposes. In this work, Bilateral Image Filter is evaluated in the field of satellite imaging for the noise removal process. The evaluated filter receives two parameters from the user side within their predefined ranges. Here, SPEA2 algorithm takes the responsibility to optimize these parameters to reach the best noise free image results. SPEA2 algorithm was implemented in Matlab and executable files of image filter were called in Matlab environment. The results of the work were represented graphically to show the effectiveness of selected method.

In this paper, we propose the method for recognizing the degree of snowfall automatically even if most of backgrounds are covered with snow and the visibility is low by fog. When it snows heavily, fog often occurs simultaneously. Moreover, falling snow grains have low contrast to the background covered with white snow. In order to deal with these situations, the proposed method makes an input image clear by fog removal. We propose the novel fog removal method which can be applied to not only the usual scene but also the heavy snowy scene. This method can remove the influence of fog without depending on the grade of visibility because the degree of fog removal is changed dynamically. The degree of snowfall is estimated from the quantity of falling snow grains, which are extracted from the difference between the present defogged image and the background image created by the three-dimensional median. Experiments conducted under various degrees of snowfall have shown the effectiveness of the proposed method.

In this paper, we propose a new transmission model using LO norm for image fog removal. In the prior work, the bilateral filter was used to reduce the halo artifacts. However, it is only a local optimization. Hence, we observe non-zero gradients to develop the gradient smoothing method for global control. The proposed model then locates significant edges to highlight the prominent parts of an image. Experimental results show the effectiveness of the proposed model for defogging.

This paper introduces the Automated Two-Dimensional K-Means (A2DKM) algorithm, a novel unsupervised clustering technique. The proposed technique differs from the conventional clustering techniques because it eliminates the need for users to determine the number of clusters. In addition, A2DKM incorporates local and spatial information of the data into the clustering analysis. A2DKM is qualitatively and quantitatively compared with the conventional clustering algorithms, namely, the K-Means (KM), Fuzzy C-Means (FCM), Moving K-Means (MKM), and Adaptive Fuzzy K-Means (AFKM) algorithms. The A2DKM outperforms these algorithms by producing more homogeneous segmentation results.

K-means algorithm and ant clustering algorithm are all traditional algorithms. The two algorithms can complement each other. The combination of two algorithms will improve clustering’s accuracy and speed up algorithm’ convergence. Tests prove hybrid clustering algorithm is more effective than each above-mentioned algorithm. Especially, the new algorithm has good results in image segmentation.

In medical image processing, interactive image segmentation is an important part, because it can obtain accurate segment results with less human effort compared with manual scribing. We proposed an improved algorithm of maximal similarity based region merging, compared with the algorithm proposed in our algorithm use SLIC superpixels segmentation to obtain presegmented regions, using SLIC superpixles, it is easy to control the number of presegmentation regions. We also introduce the texture features differeces while rigion merging, so we can obtain the accuracy of similarity measurement. Experimental results show that our algorithm can obtain comparable results.

ion Level set methods are a popular way to solve the image segmentation problem. The solution contour is found by solving an optimization problem where a cost functional is minimized. Gradient descent methods are often used to solve this optimization problem since they are very easy to implement and applicable to general nonconvex functionals. They are, however, sensitive to local minima and often display slow convergence. Traditionally, cost functionals have been modified to avoid these problems. In this paper, we instead propose using two modified gradient descent methods, one using a momentum term and one based on resilient propagation. These methods are commonly used in the machine learning community. In a series of 2- D/3-D-experiments using real and synthetic data with ground truth, the modifications are shown to reduce the sensitivity for local optima and to increase the convergence rate. The parameter sensitivity is also investigated. The proposed methods are very simple modifications of the basic method, and are directly compatible with any type of level set implementation. Downloadable reference code with examples is available online.

Security of secret data has been a major issue of concern from ancient time. Steganography and cryptography are the two techniques which are used to reduce the security threat. Cryptography is an art of converting secret message in other than human readable form. Steganography is an art of hiding the existence of secret message. These techniques are required to protect the data theft over rapidly growing network. To achieve this there is a need of such a system which is very less susceptible to human visual system. In this paper a new technique is going to be introducing for data transmission over an unsecure channel. In this paper secret data is compressed first using LZW algorithm before embedding it behind any cover media. Data is compressed to reduce its size. After compression data encryption is performed to increase the security. Encryption is performed with the help of a key which make it difficult to get the secret message even if the existence of the secret message is reveled. Now the edge of secret message is detected by using canny edge detector and then embedded secret data is stored there with the help of a hash function. Proposed technique is implemented in MATLAB and key strength of this project is its huge data hiding capacity and least distortion in Stego image. This technique is applied over various images and the results show least distortion in altered image.