Showing papers on "Standard test image published in 2009"

Object recognition and full pose registration from a single image for robotic manipulation

Abstract: Robust perception is a vital capability for robotic manipulation in unstructured scenes. In this context, full pose estimation of relevant objects in a scene is a critical step towards the introduction of robots into household environments. In this paper, we present an approach for building metric 3D models of objects using local descriptors from several images. Each model is optimized to fit a set of calibrated training images, thus obtaining the best possible alignment between the 3D model and the real object. Given a new test image, we match the local descriptors to our stored models online, using a novel combination of the RANSAC and Mean Shift algorithms to register multiple instances of each object. A robust initialization step allows for arbitrary rotation, translation and scaling of objects in the test images. The resulting system provides markerless 6-DOF pose estimation for complex objects in cluttered scenes. We provide experimental results demonstrating orientation and translation accuracy, as well a physical implementation of the pose output being used by an autonomous robot to perform grasping in highly cluttered scenes.

Face recognition with contiguous occlusion using markov random fields

Abstract: Partially occluded faces are common in many applications of face recognition While algorithms based on sparse representation have demonstrated promising results, they achieve their best performance on occlusions that are not spatially correlated (ie random pixel corruption) We show that such sparsity-based algorithms can be significantly improved by harnessing prior knowledge about the pixel error distribution We show how a Markov Random Field model for spatial continuity of the occlusion can be integrated into the computation of a sparse representation of the test image with respect to the training images Our algorithm efficiently and reliably identifies the corrupted regions and excludes them from the sparse representation Extensive experiments on both laboratory and real-world datasets show that our algorithm tolerates much larger fractions and varieties of occlusion than current state-of-the-art algorithms

A no-reference sharpness metric sensitive to blur and noise

Abstract: A no-reference objective sharpness metric detecting both blur and noise is proposed in this paper. This metric is based on the local gradients of the image and does not require any edge detection. Its value drops either when the test image becomes blurred or corrupted by random noise. It can be thought of as an indicator of the signal to noise ratio of the image. Experiments using synthetic, natural, and compressed images are presented to demonstrate the effectiveness and robustness of this metric. Its statistical properties are also provided.

A compressive sensing approach for expression-invariant face recognition

Abstract: We propose a novel technique based on compressive sensing for expression invariant face recognition. We view the different images of the same subject as an ensemble of intercorrelated signals and assume that changes due to variation in expressions are sparse with respect to the whole image. We exploit this sparsity using distributed compressive sensing theory, which enables us to grossly represent the training images of a given subject by only two feature images: one that captures the holistic (common) features of the face, and the other that captures the different expressions in all training samples. We show that a new test image of a subject can be fairly well approximated using only the two feature images from the same subject. Hence we can drastically reduce the storage space and operational dimensionality by keeping only these two feature images or their random measurements. Based on this, we design an efficient expression invariant classifier. Furthermore, we show that substantially low dimensional versions of the training features, such as ones extracted from critically downsampled training images, or low dimensional random projection of original feature images, still have sufficient information for good classification. Extensive experiments with publically available databases show that, on average, our approach performs better than the state of the art despite using only such super compact feature representation.

Generation of digital phantoms of cell nuclei and simulation of image formation in 3D image cytometry.

Abstract: Image cytometry still faces the problem of the quality of cell image analysis results. Degradations caused by cell preparation, optics and electronics considerably affect most 2D and 3D cell image data acquired using optical microscopy. That is why image processing algorithms applied to these data typically offer imprecise and unreliable results. We have created a toolbox that can generate 3D digital phantoms of specific cellular components along with their corresponding images degraded by specific optics and electronics. The user can then apply image analysis methods to such simulated image data. The analysis results can be compared with ground truth derived from input object digital phantoms. In this way, image analysis methods can be compared to each other and their quality can be computed. We have also evaluated the plausibility of the synthetic images, measured by their similarity to real image data.

Perfecting of digital image capture parameters within acquisition devices using face detection

Abstract: Within a digital acquisition device, acquisition parameters of a digital image are perfected as part of an image capture process using face detection within said captured image to achieve one or more desired image acquisition parameters. Default values are determined of one or more image attributes of at least some portion of the digital image. Values of one or more camera acquisition parameters are determined. Groups of pixels are identified that correspond to an image of a face within the digitally-captured image. Corresponding image attributes to the groups of pixels are determined. One or more default image attribute values are compared with one or more captured image attribute values based upon analysis of the image of the face. Camera acquisition parameters are then adjusted corresponding to adjusting the image attribute values.

Active skeleton for non-rigid object detection

Abstract: We present a shape-based algorithm for detecting and recognizing non-rigid objects from natural images The existing literature in this domain often cannot model the objects very well In this paper, we use the skeleton (medial axis) information to capture the main structure of an object, which has the particular advantage in modeling articulation and non-rigid deformation Given a set of training samples, a tree-union structure is learned on the extracted skeletons to model the variation in configuration Each branch on the skeleton is associated with a few part-based templates, modeling the object boundary information We then apply sum-and-max algorithm to perform rapid object detection by matching the skeleton-based active template to the edge map extracted from a test image The algorithm reports the detection result by a composition of the local maximum responses Compared with the alternatives on this topic, our algorithm requires less training samples It is simple, yet efficient and effective We show encouraging results on two widely used benchmark image sets: the Weizmann horse dataset [7] and the ETHZ dataset [16]

Distance metric learning from uncertain side information with application to automated photo tagging

Abstract: Automated photo tagging is essential to make massive unlabeled photos searchable by text search engines. Conventional image annotation approaches, though working reasonably well on small testbeds, are either computationally expensive or inaccurate when dealing with large-scale photo tagging. Recently, with the popularity of social networking websites, we observe a massive number of user-tagged images, referred to as "social images", that are available on the web. Unlike traditional web images, social images often contain tags and other user-generated content, which offer a new opportunity to resolve some long-standing challenges in multimedia. In this work, we aim to address the challenge of large-scale automated photo tagging by exploring the social images. We present a retrieval based approach for automated photo tagging. To tag a test image, the proposed approach first retrieves k social images that share the largest visual similarity with the test image. The tags of the test image are then derived based on the tagging of the similar images. Due to the well-known semantic gap issue, a regular Euclidean distance-based retrieval method often fails to find semantically relevant images. To address the challenge of semantic gap, we propose a novel probabilistic distance metric learning scheme that (1) automatically derives constraints from the uncertain side information, and (2) efficiently learns a distance metric from the derived constraints. We apply the proposed technique to automated photo tagging tasks based on a social image testbed with over 200,000 images crawled from Flickr. Encouraging results show that the proposed technique is effective and promising for automated photo tagging.

Organizing images by correlating faces

Abstract: Methods and systems are presented for organizing images. In one aspect, a method can include generating a correlation value indicating a likelihood that a face included in a test image corresponds to a face associated with a base image, determining that a correlation threshold exceeds the correlation value and that the correlation value exceeds a non-correlation threshold, generating a similarity score based on one or more exposure values and one or more color distribution values corresponding to the test image and the base image, combining the similarity score with the correlation value to generate a weighted correlation value, and determining that the test image and the base image are correlated when the weighted correlation value exceeds the correlation threshold.

Volterrafaces: Discriminant analysis using Volterra kernels

Abstract: In this paper we present a novel face classification system where we represent face images as a spatial arrangement of image patches, and seek a smooth nonlinear functional mapping for the corresponding patches such that in the range space, patches of the same face are close to one another, while patches from different faces are far apart, in L2 sense. We accomplish this using Volterra kernels, which can generate successively better approximations to any smooth nonlinear functional. During learning, for each set of corresponding patches we recover a Volterra kernel by minimizing a goodness functional defined over the range space of the sought functional. We show that for our definition of the goodness functional, which minimizes the ratio between intraclass distances and interclass distances, the problem of generating Volterra approximations, to any order, can be posed as a generalized eigenvalue problem. During testing, each patch from the test image that is classified independently, casts a vote towards image classification and the class with the maximum votes is chosen as the winner. We demonstrate the effectiveness of the proposed technique in recognizing faces by extensive experiments on Yale, CMU PIE and Extended Yale B benchmark face datasets and show that our technique consistently outperforms the state-of-the-art in learning based face discrimination.

Method and apparatus for processing 3d video image

Abstract: A 3D video image processing method including: acquiring three-dimensional (3D) format information of a video image generated from video data to determine a 3D format of the video image; generating, from a first graphic image, a second graphic image corresponding to the determined 3D format of the video image using the 3D format information, the first graphic image being generated from graphic data; and overlaying the video image with the second graphic image.

Efficient multi-label ranking for multi-class learning: Application to object recognition

Abstract: Multi-label learning is useful in visual object recognition when several objects are present in an image Conventional approaches implement multi-label learning as a set of binary classification problems, but they suffer from imbalanced data distributions when the number of classes is large In this paper, we address multi-label learning with many classes via a ranking approach, termed multi-label ranking Given a test image, the proposed scheme aims to order all the object classes such that the relevant classes are ranked higher than the irrelevant ones We present an efficient algorithm for multi-label ranking based on the idea of block coordinate descent The proposed algorithm is applied to visual object recognition Empirical results on the PASCAL VOC 2006 and 2007 data sets show promising results in comparison to the state-of-the-art algorithms for multi-label learning

Device and method for estimating whether an image is blurred

Abstract: The present invention is directed to a method for detecting or predicting (302, 602) whether a test image is blurred. In one embodiment, the method includes extracting a training statistical signature (366) that is based on a plurality of data features (362, 364) from a training image set (14, 16), the training image set (14, 16) including a sharp image (14) and a blurry image (16); training a classifier (368) to discriminate between the sharp image (14) and the blurry image (16) based on the training statistical signature; and applying (302, 602) the trained classifier to a test image that is not included in the training image set (14, 16) to predict whether the test image is sharp (18) or blurry (20). The step of extracting can include measuring one or more statistical moments (576, 776) for various levels (L0-L5), estimating a covariance (577, 777) between adjacent levels (L0-L5), and/or extracting various metadata features (364, 664) from the images (14, 16). The step of training (300, 600) can include training a non-linear support vector machine (300) or a linear discriminant analysis (600) on the training statistical signature of the training image set (14, 16).

Image Scaling Comparison Using Universal Image Quality Index

Abstract: Image interpolation has many applications in computer vision, image processing and biomedical applications. Resampling is required for discrete image manipulations, such as geometric alignment and registration, to improve image quality on display devices or in the field of lossy image compression wherein some pixels are discarded during the encoding process and must be regenerated from the remaining information for decoding. The comparison is done for different interpolation techniques such as nearest neighbor, bilinear and bicubic interpolation and the comparison is done for different interpolation schemes using universal image quality index. In this paper an attempt is made to highlight the universal quality index by comparing with error measures such as MSE and PSNR.

Facial age estimation by multilinear subspace analysis

Abstract: Automatic estimation of human facial age is an interesting yet challenging topic appearing in recent years. Since different people might age in different ways, solving the problem of age estimation involves two semantic labels: identity and age. In this paper, aging face images are organized in a third-order tensor according to both identity and age. Due to the difficulty in data collection, the aging pattern for each person in the training set is always incomplete. Therefore, the tensor contains a large amount of missing values. Through a series of multilinear subspace analysis algorithms operating on tensor with missing values, the aging pattern contained in the training aging images can be iteratively learned and be used to predict the age of a given test image. In the experiment, the proposed method not only outperforms the existing algorithms, but also exceeds the human ability in age estimation.

Method and system for automated x-ray inspection of objects

Abstract: An anomaly detection method and system for comparing a scanned object to an idealized object is provided. The anomaly detection method includes generating a three-dimensional reference model of the idealized object. The anomaly detection method further includes acquiring at least one two-dimensional inspection test image of the scanned object. The anamoly detection method also includes determining a two-dimensional reference image from the three-dimensional reference model using multiple pose parameters, wherein the two-dimensional reference image corresponds to the same view of the three-dimensional reference model of the idealized object as the view of the two-dimensional inspection test image of the scanned object. The anamoly detection method further includes identifying one or more defects in the inspection test image via automated defect recognition technique.

Natural image utility assessment using image contours

Abstract: In the quality assessment task, observers evaluate a natural image based on its perceptual resemblance to a reference. For the utility assessment task, observers evaluate the usefulness of a natural image as a surrogate for a reference. Humans generally use the information captured by an imaging system and tolerate distortions as long as the underlying task is performed reliably. Conventional notions of perceived quality cannot generally predict the perceived utility of a natural image. This paper examines variations to basic components of a recently introduced utility assessment algorithm that compares the contours of a reference and test image, referred to as the natural image contour evaluation (NICE), in terms of their capability to improve the prediction of perceived utility scores. Results show that classical edge-detection algorithms incorporated into NICE provide statistically equivalent performance to other, more complex edge-detection algorithms.

A robust image watermarking algorithm using SVR detection

Abstract: Geometric distortion is known as one of the most difficult attacks to resist. Geometric distortion desynchronizes the location of the watermark and hence causes incorrect watermark detection. According to the Support Vector Regression (SVR), a new image watermarking detection algorithm against geometric attacks is proposed in this paper, in which the steady Pseudo-Zernike moments and Krawtchouk moments are utilized. The host image is firstly transformed from rectangular coordinates to polar coordinates, and the Pseudo-Zernike moments of the host image are computed. Then some low-order Pseudo-Zernike moments are selected, and the digital watermark is embedded into the cover image by quantizing the magnitudes of the selected Pseudo-Zernike moments. The main steps of watermark detecting procedure include: (i) some low-order Krawtchouk moments of the image are calculated, which are taken as the eigenvectors; (ii) the geometric transformation parameters are regarded as the training objective, the appropriate kernel function is selected for training, and a SVR training model can be obtained; (iii) the Krawtchouk moments of test image are selected as input vector, the actual output (geometric transformation parameters) is predicted by using the well trained SVR, and the geometric correction is performed on the test image by using the obtained geometric transformation parameters; (iv) the digital watermark is extracted from the corrected test image. Experimental results show that the proposed watermarking detection algorithm is not only robust against common signal processing such as filtering, sharpening, noise adding, and JPEG compression etc., but also robust against the geometric attacks such as rotation, translation, scaling, cropping and combination attacks, etc.

Off-Line Persian Signature Identification and Verification Based on Image Registration and Fusion

Abstract: Signature verification and Identification has great importance for authentication purpose. Persian signatures are different from other signature types because people usually do not use text in it and they draw a shape as their signature, therefore, a different approach should be considered to process such signatures. In this paper, a method for off-line Persian signature identification and verification is proposed that is based on Image Registration, DWT (Discrete Wavelet Transform) and Image Fusion. Training signatures of each person are registered to overcome shift and scale problem. To extract features, at first, DWT is used to access details of signature; then several registered instances of each person signatures are fused together to generate reference pattern of person's signatures. In the classification phase, Euclidean distance between the test image and each pattern is used in different sub-bands. Experimental results confirmed the effectiveness of the proposed method. However, the proposed method has been tested on Persian signature database but we believe it can be extended for other languages.

Image Utility Assessment and a Relationship with Image Quality Assessment

Abstract: Present quality assessment (QA) algorithms aim to generate scores for natural images consistent with subjective scores for the quality assessment task. For the quality assessment task, human observers evaluate a natural image based on its perceptual resemblance to a reference. Natural images communicate useful information to humans, and this paper investigates the utility assessment task, where human observers evaluate the usefulness of a natural image as a surrogate for a reference. Current QA algorithms implicitly assess utility insofar as an image that exhibits strong perceptual resemblance to a reference is also of high utility. However, a perceived quality score is not a proxy for a perceived utility score: a decrease in perceived quality may not affect the perceived utility. Two experiments are conducted to investigate the relationship between the quality assessment and utility assessment tasks. The results from these experiments provide evidence that any algorithm optimized to predict perceived quality scores cannot immediately predict perceived utility scores. Several QA algorithms are evaluated in terms of their ability to predict subjective scores for the quality and utility assessment tasks. Among the QA algorithms evaluated, the visual information fidelity (VIF) criterion, which is frequently reported to provide the highest correlation with perceived quality, predicted both perceived quality and utility scores reasonably. The consistent performance of VIF for both the tasks raised suspicions in light of the evidence from the psychophysical experiments. A thorough analysis of VIF revealed that it artificially emphasizes evaluations at finer image scales (i.e., higher spatial frequencies) over those at coarser image scales (i.e., lower spatial frequencies). A modified implementation of VIF, denoted VIF*, is presented that provides statistically significant improvement over VIF for the quality assessment task and statistically worse performance for the utility assessment task. A novel utility assessment algorithm, referred to as the natural image contour evaluation (NICE), is introduced that conducts a comparison of the contours of a test image to those of a reference image across multiple image scales to score the test image. NICE demonstrates a viable departure from traditional QA algorithms that incorporate energy-based approaches and is capable of predicting perceived utility scores.

A new reference-free image quality index for blur estimation in the frequency domain

Abstract: A new reference-free image quality index based on spectral analysis is proposed. The main idea is based on exploiting the limitations of the Human Visual System (HVS) in blur detection,. The proposed method consists of adding blur to the test image and measuring its impact. The impact is measured using radial analysis in the frequency domain. The efficiency of the proposed method is tested objectively by comparing it to some well known algorithms and in terms of correlation with subjective scores.

Estimating human pose from occluded images

Abstract: We address the problem of recovering 3D human pose from single 2D images, in which the pose estimation problem is formulated as a direct nonlinear regression from image observation to 3D joint positions. One key issue that has not been addressed in the literature is how to estimate 3D pose when humans in the scenes are partially or heavily occluded. When occlusions occur, features extracted from image observations (e.g., silhouettes-based shape features, histogram of oriented gradient, etc.) are seriously corrupted, and consequently the regressor (trained on un-occluded images) is unable to estimate pose states correctly. In this paper, we present a method that is capable of handling occlusions using sparse signal representations, in which each test sample is represented as a compact linear combination of training samples. The sparsest solution can then be efficiently obtained by solving a convex optimization problem with certain norms (such as l1-norm). The corrupted test image can be recovered with a sparse linear combination of un-occluded training images which can then be used for estimating human pose correctly (as if no occlusions exist). We also show that the proposed approach implicitly performs relevant feature selection with un-occluded test images. Experimental results on synthetic and real data sets bear out our theory that with sparse representation 3D human pose can be robustly estimated when humans are partially or heavily occluded in the scenes.

Comparative Analysis of Image Compression Techniques: A Case Study on Medical Images

Abstract: Ultrasound, computed tomography (CT), magnetic resonance imaging (MRI) medical imaging produce human body pictures in digital form. These medical applications have already been integrated into mobile devices and are being used by medical personnel in treatment centers, for retrieving and examining patient data and medical images. Storage and transmission are key issues in such platforms, due to the significant image file sizes. Wavelet transform has been considered to be a highly efficient technique of image compression resulting in both lossless and lossy compression of images with great accuracy, enabling its use on medical images. On the other hand, in some areas in medicine, it may be sufficient to maintain high image quality only in the region of interest i.e. in diagnostically important regions. This paper proposes a framework for ROI based compression of medical images using wavelet based compression techniques (i.e. JPEG2000 and SPIHT). Results are analyzed by conducting the experiments on a number of medical images by taking different region of interests. The performance is evaluated using various image quality metrics like PSNR, SSIM and Correlation.

Pose Invariant Palmprint Recognition

Abstract: A palmprint based authentication system that can work with a multi-purpose camera in uncontrolled circumstances, such as those mounted on a laptop, mobile device or those for surveillance, can dramatically increase the applicability of such a system. However, the performance of existing techniques for palmprint authentication fall considerably, when the camera is not aligned with the surface of the palm. The problems arise primarily due to variations in appearance introduced due to varying pose, but is compounded by specularity of the skin and blur due to motion and focus. In this paper, we propose a method to deal with variations in pose in unconstrained palmprint imaging. The method can robustly estimate and correct variations in pose, and compute a similarity measure between the corrected test image and a reference image. Experimental results on a set of 100 user's palms captured at varying poses show a reduction in Equal Error Eate from 22.4% to 8.7%.

Diagnostic image processing with automatic self image quality validation

Abstract: A method for medical diagnostic image processing obtains digital image data for a diagnostic image and extracts one or more image features from the image data, obtaining one or more image properties from the one or more extracted features. An image quality aim is obtained for rendered image appearance according to one or more stored viewer preferences. Rendering parameters are generated according to the obtained image quality aim and the one or more obtained image properties. The image is rendered according to the generated rendering parameters and the rendering validated against the selected image quality aim.

Nondestructive inspection method and system

Abstract: A method and system for nondestructively detecting and quantifying material anomalies within materials, including composite articles. The method entails performing a three-dimensional imaging scan technique, such as a computed tomography scan, of the material and a reference standard such that a test image of the material and a reference image of the reference standard appear in a plurality of two-dimensional scan views generated by the scan technique. The reference images are located in the scan views and normalized to determine at least an average value of the pixel data for the reference images. Values of pixel data of the test image are determined in each scan view, and then compared to the pixel data of the reference images to detect the presence of an anomaly in the test images. The detected anomaly in at least one of the test images of the scan views is then compared to a requirement standard for the material.

Image processing apparatus, image processing method, and computer program product

Abstract: An image processing apparatus, method and computer program that controls so that an image of image data is displayed on a display unit, controls so that a region of interest is indicated on the displayed image to acquire image data of the region of interest, generates an extraction region extracted from the image data by using each of the image segmentation algorithms to acquire the image data of the extraction region, calculates similarity by comparing the image data of the extraction region with the image data of the region of interest to select the image segmentation algorithm having highest similarity, and outputs image data extracted using the selected image segmentation algorithm to the display unit.

Bayesian prior models for vehicle make and model recognition

Abstract: Automatic vehicle type recognition (make and model) is very useful in secure access and traffic monitoring applications. It compliments the number plate recognition systems by providing a higher level of security against fraudulent use of number plates in traffic crimes. In this paper we present a simple but powerful probabilistic framework for vehicle type recognition that requires just a single representative car image in the database to recognize any incoming test image exhibiting strong appearance variations, as expected in outdoor image capture e.g. illumination, scale etc. We propose to use a new feature description, local energy based shape histogram 'LESH', in this problem that encodes the underlying shape and is invariant to illumination and other appearance variations such as scale, perspective distortions and color. Our method achieves high accuracy (above 94%) as compared to the state of the art previous approaches on a standard benchmark car dataset. It provides a posterior over possible vehicle type matches which is especially attractive and very useful in practical traffic monitoring and/or surveillance video search (for a specific vehicle type) applications.

Adaptive technique for human face detection using HSV color space and neural networks

Abstract: Human face detection has a wide area of applications in image processing. A new face detection technique is presented in this paper. The technique combines two methods for face detection to achieve better detection rates. The two methods are skin face detection using HSV color space and the back propagation neural networks. In the first module of the technique, we proposed a fast algorithm for detecting human faces in color images. The algorithm uses color histogram of HSV color space to detect skin regions. In the second module of the technique, the neural network only examined the face candidate regions instead of performing exhaustive search in every part of the test image, so we reduced the search space. In experiments on images, our technique has achieved high detection rates and low false positives due to the application of the new algorithm.

Real-time sidescan simulator and applications

Abstract: New generations of sonars appeared in the last decade. The major interest in SAS systems and high frequency sonars is in the improvement of the sonar resolution and the reduction of noise level. Sonar images are distance-images but at high resolution they tends to appear visually as optical images. Usually the algorithms developed for sidescan were specific for sonar images due to the poor resolution essentially. With high resolution sonars, algorithms developed in the image processing field for natural images became applicable. In this paper we present a real-time and realistic sidescan simulator, and test image-based classification algorithms (such as PCA and eigenface algorithms) with synthetic images in order to characterize the precision necessary for these image-based algorithm to work.