scispace - formally typeset
Search or ask a question

Showing papers on "Standard test image published in 2005"


Journal ArticleDOI
Wei Tong1
01 Nov 2005-Strain
TL;DR: In this article, the performance of four digital image correlation criteria widely used in strain mapping applications has been critically examined using three sets of digital images with various whole-field deformation characteristics.
Abstract: The performance of four digital image correlation criteria widely used in strain mapping applications has been critically examined using three sets of digital images with various whole-field deformation characteristics. The deformed images in these image sets are digitally modified to simulate the less-than-ideal image acquisition conditions in an actual experiment, such as variable brightness, contrast, uneven local lighting and blurring. The relative robustness, computational cost and reliability of each criterion are assessed for precision strain mapping applications. Recommen- dations are given for selecting a proper image correlation criterion to efficiently extract reliable deformation data from a given set of digital images.

244 citations


Proceedings ArticleDOI
17 Oct 2005
TL;DR: An algorithm for a non-negative 3D tensor factorization for establishing a local parts feature decomposition from an object class of images shows a superior decomposition to what an NMF can provide on all fronts.
Abstract: We introduce an algorithm for a non-negative 3D tensor factorization for the purpose of establishing a local parts feature decomposition from an object class of images. In the past, such a decomposition was obtained using non-negative matrix factorization (NMF) where images were vectorized before being factored by NMF. A tensor factorization (NTF) on the other hand preserves the 2D representations of images and provides a unique factorization (unlike NMF which is not unique). The resulting "factors" from the NTF factorization are both sparse (like with NMF) but also separable allowing efficient convolution with the test image. Results show a superior decomposition to what an NMF can provide on all fronts - degree of sparsity, lack of ghost residue due to invariant parts and efficiency of coding of around an order of magnitude better. Experiments on using the local parts decomposition for face detection using SVM and Adaboost classifiers demonstrate that the recovered features are discriminatory and highly effective for classification.

204 citations


Proceedings ArticleDOI
20 Jun 2005
TL;DR: This work introduces a nonlinear, multifactor model that generalizes ICA and demonstrates that the statistical regularities learned by MICA capture information that, in conjunction with the multilinear projection algorithm, improves automatic face recognition.
Abstract: Independent components analysis (ICA) maximizes the statistical independence of the representational components of a training image ensemble, but it cannot distinguish between the different factors, or modes, inherent to image formation, including scene structure, illumination, and imaging. We introduce a nonlinear, multifactor model that generalizes ICA. Our multilinear ICA (MICA) model of image ensembles learns the statistically independent components of multiple factors. Whereas ICA employs linear (matrix) algebra, MICA exploits multilinear (tensor) algebra. We furthermore introduce a multilinear projection algorithm which projects an unlabeled test image into the N constituent mode spaces to simultaneously infer its mode labels. In the context of facial image ensembles, where the mode labels are person, viewpoint, illumination, expression, etc., we demonstrate that the statistical regularities learned by MICA capture information that, in conjunction with our multilinear projection algorithm, improves automatic face recognition.

201 citations


Journal ArticleDOI
TL;DR: This work presents a novel texture and shape priors based method for kidney segmentation in ultrasound (US) images that is demonstrated through experimental results on both natural images and US data compared with other image segmentation methods and manual segmentation.
Abstract: This work presents a novel texture and shape priors based method for kidney segmentation in ultrasound (US) images. Texture features are extracted by applying a bank of Gabor filters on test images through a two-sided convolution strategy. The texture model is constructed via estimating the parameters of a set of mixtures of half-planed Gaussians using the expectation-maximization method. Through this texture model, the texture similarities of areas around the segmenting curve are measured in the inside and outside regions, respectively. We also present an iterative segmentation framework to combine the texture measures into the parametric shape model proposed by Leventon and Faugeras. Segmentation is implemented by calculating the parameters of the shape model to minimize a novel energy function. The goal of this energy function is to partition the test image into two regions, the inside one with high texture similarity and low texture variance, and the outside one with high texture variance. The effectiveness of this method is demonstrated through experimental results on both natural images and US data compared with other image segmentation methods and manual segmentation.

195 citations


Proceedings ArticleDOI
06 Jul 2005
TL;DR: A general blind image steganalysis system is proposed, in which the statistical moments of characteristic functions of the prediction-error image, the test image, and their wavelet subbands are selected as features.
Abstract: In this paper, a general blind image steganalysis system is proposed, in which the statistical moments of characteristic functions of the prediction-error image, the test image, and their wavelet subbands are selected as features. Artificial neural network is utilized as the classifier. The performance of the proposed steganalysis system is significantly superior to the prior arts.

161 citations


Patent
30 Mar 2005
TL;DR: In this paper, a method of measuring performance parameters of an imaging device (120, 160) is disclosed, which maintains a test pattern image (1005) consisting alignment features and image analysis features.
Abstract: A method (1000) of measuring performance parameters of an imaging device (120, 160) is disclosed. The method (1000) maintains a test pattern image (1005), the test pattern image (1005) comprising alignment features and image analysis features. A test chart (110, 170) containing a representation of the test pattern image is next imaged using the imaging device (120, 160) to form a second image (1010). The test pattern image (1005) and the second image (1010) are then registered using region based matching (1035) operating on the alignment features. Finally, the performance parameters are measured by analysing (1060) the image analysis features.

152 citations


Proceedings ArticleDOI
15 Aug 2005
TL;DR: The paper proposes methods to use a hierarchy defined on the annotation words derived from a text ontology to improve automatic image annotation and retrieval and demonstrates improvements in the annotation performance of translation models.
Abstract: Automatic image annotation is the task of automatically assigning words to an image that describe the content of the image. Machine learning approaches have been explored to model the association between words and images from an annotated set of images and generate annotations for a test image. The paper proposes methods to use a hierarchy defined on the annotation words derived from a text ontology to improve automatic image annotation and retrieval. Specifically, the hierarchy is used in the context of generating a visual vocabulary for representing images and as a framework for the proposed hierarchical classification approach for automatic image annotation. The effect of using the hierarchy in generating the visual vocabulary is demonstrated by improvements in the annotation performance of translation models. In addition to performance improvements, hierarchical classification approaches yield well to constructing multimedia ontologies.

120 citations


Proceedings ArticleDOI
TL;DR: A frequency domain POCS algorithm for the canonical problem of super-resolution (SR) image synthesis is proposed, structured to accommodate rotations of the source relative to the imaging device.
Abstract: Optical imaging systems are often limited in resolution, not by the imaging optics, but by the light intensity sensors on the image formation plane. When the sensor size is larger than the optical spot size, the effect is to smooth the image with a rectangular convolving kernel with one sample at each non-overlapping kernel position, resulting in aliasing. In some such imaging systems, there is the possibility of collecting multiple images of the same scene. The process of reconstructing a de-aliased high-resolution image from multiple images of this kind is referred to as “super-resolution image reconstruction.” We apply the POCS method to this problem in the frequency domain. Generally, frequency domain methods have been used when component images were related by subpixel shifts only, because rotations of a sampled image do not correspond to a simple operation in the frequency domain. This algorithm is structured to accommodate rotations of the source relative to the imaging device, which we believe helps in producing a well-conditioned image synthesis problem. A finely sampled test image is repeatedly resampled to align with each observed image. Once aligned, the test and observed images are readily related in the frequency domain and a projection operation is defined.

117 citations


Proceedings ArticleDOI
17 Oct 2005
TL;DR: This work reduces the cost of searching for matches between video frames by adaptively identifying key frames based on the amount of image-to-image overlap, and substantially reduces the time required to estimate the image orientations with minimal loss of accuracy.
Abstract: We present an automatic and efficient method to register and stitch thousands of video frames into a large panoramic mosaic. Our method preserves the robustness and accuracy of image stitchers that match all pairs of images while utilizing the ordering information provided by video. We reduce the cost of searching for matches between video frames by adaptively identifying key frames based on the amount of image-to-image overlap. Key frames are matched to all other key frames, but intermediate video frames are only matched to temporally neighboring key frames and intermediate frames. Image orientations can be estimated from this sparse set of matches in time quadratic to cubic in the number of key frames but only linear in the number of intermediate frames. Additionally, the matches between pairs of images are compressed by replacing measurements within small windows in the image with a single representative measurement. We show that this approach substantially reduces the time required to estimate the image orientations with minimal loss of accuracy. Finally, we demonstrate both the efficiency and quality of our results by registering several long video sequences

109 citations


Proceedings ArticleDOI
01 Jan 2005
TL;DR: A method capable of recognising one of N objects in log(N) time, which preserves all the strengths of local affine region methods – robustness to background clutter, occlusion, and large changes of viewpoints.
Abstract: Realistic approaches to large scale object recognition, i.e. for detection and localisation of hundreds or more objects, must support sub-linear time indexing. In the paper, we propose a method capable of recognising one of N objects in log(N) time. The ”visual memory” is organised as a binary decision tree that is built to minimise average time to decision. Leaves of the tree represent a few local image areas, and each non-terminal node is associated with a ’weak classifier’. In the recognition phase, a single invariant measurement decides in which subtree a corresponding image area is sought. The method preserves all the strengths of local affine region methods – robustness to background clutter, occlusion, and large changes of viewpoints. Experimentally we show that it supports near real-time recognition of hundreds of objects with state-of-the-art recognition rates. After the test image is processed (in a second on a current PCs), the recognition via indexing into the visual memory requires milliseconds.

92 citations


Journal ArticleDOI
TL;DR: A method of regularization by minimizing Csiszar's I-divergence measure is derived that produces an image that both is consistent with the known data and extrapolates additional detail based on constraints on the magnitude of the image.
Abstract: For optical coherence tomography (OCT), ultrasound, synthetic-aperture radar, and other coherent ranging methods, speckle can cause spurious detail that detracts from the utility of the image. It is a problem inherent to imaging densely scattering objects with limited bandwidth. Using a method of regularization by minimizing Csiszar's I-divergence measure, we derive a method of speckle minimization that produces an image that both is consistent with the known data and extrapolates additional detail based on constraints on the magnitude of the image. This method is demonstrated on a test image and on an OCT image of a Xenopus laevis tadpole. © 2005 Optical Society of America OCIS codes: 100.3010, 110.4500. 2 (OCT), ultrasound, and synthetic-aperture radar. Natural objects usually contain detail at all scales, but most coherent ranging instruments can probe these details in only a narrow bandwidth of frequencies. The re- sult is speckle, which is caused by the seemingly random interference of scatterers within a resolution cell of the in- strument. Speckle randomizes the amplitudes of points in the image, which obscures fine features but generally al- lows coarse features to be resolved. We propose and dem- onstrate a method of regularization that removes these small-scale amplitude changes without smoothing the im- age. This is achieved by constraining the resulting recon- structed image to be consistent in a least-squares sense with the known data, while simultaneously utilizing Csiszar's I-divergence measure 3,4 to constrain and regu- larize the amplitudes of the reconstructed data. By ensur- ing that the resulting image is consistent with the known data, the detail is retained, while additional bandwidth is extrapolated to make the amplitudes of the image consis- tent with the regularization. Most natural objects, especially those biological in ori- gin, have detail at length scales from the macroscopic to the atomic. Unfortunately, because of instrument limita- tions and the attenutation of radiation by the sample, only a finite signal bandwidth can be utilized for useful resolution from the sample. The phenomenon of speckle in these imaging modalities is caused by the interference between scatterers that are too small and too close to- gether to be individually resolved. What is observed is a seemingly random modulation of the amplitude of the im- age as a result of the interference of scattered waves from these unresolved scatterers. Unfortunately, it is difficult in practice to differentiate between the features that are caused by the speckle modulation and the features of interest. Because scatter- ers are randomly placed in most natural objects, the coarse detail tends to be present in almost any frequency bandwidth that is much wider than the reciprocal of the minimum feature size one wishes to resolve. This is what enables most coherent ranging techniques to be used in practice. However, as one considers distinguishing fea- tures on the minimally resolvable scale of the instrument, it is less certain whether a certain feature is due to the random modulation of speckle or to scatterers with larger sizes.

Journal ArticleDOI
TL;DR: A novel method to automatically segment subcortical structures of human brain in magnetic resonance images by using fuzzy templates that does not require specific expert definition of each structure or manual interactions during segmentation process is proposed.

Proceedings Article
01 Sep 2005
TL;DR: This paper proposes a method based on the neighbor bit planes of the image to distinguish the original images from the altered ones, the genuine ones from the doctored ones, and shows that the correlation between the bit planes as well the binary texture characteristics within the bits will differ between an original and a doctored image.
Abstract: Since extremely powerful technologies are now available to generate and process digital images, there is a concomitant need for developing techniques to distinguish the original images from the altered ones, the genuine ones from the doctored ones In this paper we focus on this problem and propose a method based on the neighbor bit planes of the image The basic idea is that, the correlation between the bit planes as well the binary texture characteristics within the bit planes will differ between an original and a doctored image This change in the intrinsic characteristics of the image can be monitored via the quantal-spatial moments of the bit planes These so-called Binary Similarity Measures are used as features in classifier design It has been shown that the linear classifiers based on BSM features can detect with satisfactory reliability most of the image doctoring executed via Photoshop tool

Patent
14 Dec 2005
TL;DR: In this article, a method for binning defects detected on a specimen is described, where the defect is assigned to a bin corresponding to the region of interest associated with the reference image, and the test image includes an image of one or more patterned features formed on the sample proximate to a defect detected on the specimen.
Abstract: Methods and systems for binning defects detected on a specimen are provided. One method includes comparing a test image to reference images. The test image includes an image of one or more patterned features formed on the specimen proximate to a defect detected on the specimen. The reference images include images of one or more patterned features associated with different regions of interest within a device being formed on the specimen. If the one or more patterned features of the test image match the one or more patterned features of one of the reference images, the method includes assigning the defect to a bin corresponding to the region of interest associated with the reference image.

Proceedings ArticleDOI
04 Apr 2005
TL;DR: With randomly selected 100 images for training and the remaining 996 images for testing, the proposed steganalysis system can steadily achieve a correct classification rate of 79% for the non-blind Spread Spectrum watermarking algorithm proposed by Cox et ai, 88% forThe blind Spread SpectrumWatermarking algorithms proposed by Piva et ao, and 91% for a generic LSB embedding method, thus indicating significant advancement in Steganalysis.
Abstract: In this paper, an effective steganalysis based on statistical moments of wavelet characteristic function is proposed. It decomposes the test image using two-level Haar wavelet transform into nine subbands (here the image itself is considered as the LL/sub 0/ subband). For each subband, the characteristic function is calculated. The first and second statistical moments of the characteristic functions from all the subbands are selected to form an 18-dimensional feature vector for steganalysis. The Bayes classifier is utilized in classification. All of the 1096 images from the CorelDraw image database are used in our extensive experimental work. With randomly selected 100 images for training and the remaining 996 images for testing, the proposed steganalysis system can steadily achieve a correct classification rate of 79% for the non-blind Spread Spectrum watermarking algorithm proposed by Cox et ai, 88% for the blind Spread Spectrum watermarking algorithm proposed by Piva et ai, and 91% for a generic LSB embedding method, thus indicating significant advancement in steganalysis.

Patent
18 Apr 2005
TL;DR: In this article, a blur checking region is determined by selecting a region in which a blur tends to clearly appear in a digital photograph image as the blur-checking region, and an image of a region corresponding to the blur checking regions is obtained as a checking image in a corrected image obtained by performing blur correction processing on the digital image.
Abstract: A blur checking region is determined by selecting a region in which a blur tends to clearly appear in a digital photograph image as the blur checking region. Then, an image of a region corresponding to the blur checking region is obtained as a checking image in a corrected image obtained by performing blur correction processing on the digital photograph image. Then, the obtained checking image is displayed in a size appropriate for the resolution of a display device.

Patent
Randy Ubillos1, Laurent Perrodin1, Dan Waylonis1, Stan Jirman1, Sarah Brody1, Mike Mages1 
04 Oct 2005
TL;DR: In this paper, a method and apparatus for managing digital images is provided, where a collection of digital images may be managed using a digital image system that displays images using groups, stacks, and versions.
Abstract: A method and apparatus for managing digital images is provided. A collection of digital images may be managed using a digital image system that displays images using groups, stacks, and versions. A group is a set of unordered digital images that may be visually represented, in a first state, using a representative image, and in a second state, by each digital image in the group. Stacks are similar to groups, except that each digital image in a stack has a rank, and each digital image in the stack is depicted in order of its rank. Versions are similar to groups, except that one or more images in the group are derived from another image in the group.

Patent
Tetsuya Takeshita1
02 May 2005
TL;DR: In this article, an electronic camera controls an imaging section to generate a test image without flash and a main image with flash, then, the electronic camera judges correlation between histogram distributions of the test image and the main image.
Abstract: An electronic camera controls an imaging section to generate a test image without flash and a main image with flash. Then, the electronic camera judges correlation between histogram distributions of the test image and the main image. If the correlation between the histogram distributions is low, the electronic camera judges that the flash illumination is uneven and performs white balance adjustments, placing greater importance on the color temperature of the flash. Conversely, if the correlation between the histogram distributions is high, the electronic camera judges that the flash illumination is uniform and performs white balance adjustments, placing greater importance on the color temperature of the main image.

Proceedings ArticleDOI
20 Jun 2005
TL;DR: This paper presents a stratified approach to both binary (outdoor-indoor) and multiple category of scene classification, which first learns mixture models for 20 basic classes of local image content based on color and texture information, and produces 20 probability density response maps indicating the likelihood that each image region was produced by each class.
Abstract: Classifying pictures into one of several semantic categories is a classical image understanding problem. In this paper, we present a stratified approach to both binary (outdoor-indoor) and multiple category of scene classification. We first learn mixture models for 20 basic classes of local image content based on color and texture information. Once trained, these models are applied to a test image, and produce 20 probability density response maps (PDRM) indicating the likelihood that each image region was produced by each class. We then extract some very simple features from those PDRMs, and use them to train a bagged LDA classifier for 10 scene categories. For this process, no explicit region segmentation or spatial context model are computed. To test this classification system, we created a labeled database of 1500 photos taken under very different environment and lighting conditions, using different cameras, and from 43 persons over 5 years. The classification rate of outdoor-indoor classification is 93.8%, and the classification rate for 10 scene categories is 90.1%. As a byproduct, local image patches can be contextually labeled into the 20 basic material classes by using loopy belief propagation (Yedidia et al., 2001) as an anisotropic filter on PDRMs, producing an image-level segmentation if desired.

Patent
Hiroshi Tojo1
13 Sep 2005
TL;DR: In this article, image data similar to image data input as a query are retrieved from an already registered image data group, and image quality of one of the retrieved image data is lower than that of the input image data.
Abstract: From an already registered image data group, image data similar to image data input as a query are retrieved (S 3190 ). When image quality of one of the retrieved image data is lower than that of the input image data, the input image data is registered in place of the one image data (S 3163 ).

Book ChapterDOI
20 Sep 2005
TL;DR: Image pre-filtering is shown to be expedient for coded image quality improvement and/or increase of compression ratio and some recommendations on how to set the compression ratio to provide quasioptimal quality of coded images are given.
Abstract: Lossy compression of noise-free and noisy images differs from each other. While in the first case image quality is decreasing with an increase of compression ratio, in the second case coding image quality evaluated with respect to a noise-free image can be improved for some range of compression ratios. This paper is devoted to the problem of lossy compression of noisy images that can take place, e.g., in compression of remote sensing data. The efficiency of several approaches to this problem is studied. Image pre-filtering is shown to be expedient for coded image quality improvement and/or increase of compression ratio. Some recommendations on how to set the compression ratio to provide quasioptimal quality of coded images are given. A novel DCT-based image compression method is briefly described and its performance is compared to JPEG and JPEG2000 with application to lossy noisy image coding.

Patent
25 May 2005
TL;DR: In this article, a centralized database for storing image correction data for imaging devices is used to process digital images, and the image processor processes the image data using the retrieved image corrections to correct the image by reducing various noise components in the image.
Abstract: An image processing system uses centralized image correction data to process digital images. The image processing system includes a centralized database for storing image correction data for imaging devices. An image processor that receives image data representing an image captured by one of the imaging devices accesses the centralized database with a key associated with the imaging device to retrieve the image correction data for the imaging device. The image processor processes the image data using the retrieved image correction data to correct the image data by reducing various noise components in the image.

Patent
21 Mar 2005
TL;DR: In this article, the thinning rate is determined such that necessary information for proper checking is maintained and unnecessary information for checking is thinned, and image correction is performed for the simple raw image data with predetermined processing parameters.
Abstract: Raw image data obtained by shooting is thinned at a thinning rate, which is determined in accordance with a shooting scene, to produce simple raw image data. The thinning rate is determined such that necessary information for proper checking is maintained and unnecessary information for checking is thinned. Image correction is performed for the simple raw image data with predetermined processing parameters, and this data is converted into a simple display image to be shown on a monitor. An operator observes the simple display image to change the processing parameters. Whenever the processing parameters are changed, the image correction is performed for the simple raw image data with the changed processing parameters to update the simple display image. For the raw image data, image correction is performed with the finally determined processing parameters.

Proceedings ArticleDOI
14 Nov 2005
TL;DR: New methods for fast, easy-to-use image color correction, with specialization toward skin tones, and fully automated estimation of facial skin color, with robustness to shadows, specularities, and blemishes are presented.
Abstract: Little prior image processing work has addressed estimation and classification of skin color in a manner that is independent of camera and illuminant. To this end, we first present new methods for 1) fast, easy-to-use image color correction, with specialization toward skin tones, and 2) fully automated estimation of facial skin color, with robustness to shadows, specularities, and blemishes. Each of these is validated independently against ground truth, and then combined with a classification method that successfully discriminates skin color across a population of people imaged with several different cameras. We also evaluate the effects of image quality and various algorithmic choices on our classification performance. We believe our methods are practical for relatively untrained operators, using inexpensive consumer equipment.

Book ChapterDOI
21 Oct 2005
TL;DR: A recent generic method for image classification based on ensemble of decision trees and random subwindows is applied and slightly adapt to achieve classification results close to the state of the art on a publicly available database of 10000 x-ray images.
Abstract: In this paper, we address a problem of biomedical image classification that involves the automatic classification of x-ray images in 57 predefined classes with large intra-class variability. To achieve that goal, we apply and slightly adapt a recent generic method for image classification based on ensemble of decision trees and random subwindows. We obtain classification results close to the state of the art on a publicly available database of 10000 x-ray images. We also provide some clues to interpret the classification of each image in terms of subwindow relevance.

Patent
05 Jul 2005
TL;DR: In this paper, face image of a character in an image represented by image data fed to a mask data creating apparatus is detected, face image areas of the detected character are respectively extracted by different two types of processing.
Abstract: When a face image of a character in an image represented by image data fed to a mask data creating apparatus is detected, face image areas of the detected character are respectively extracted by different two types of processing. An appropriate degree is calculated for each of the face image areas extracted by the two types of extraction processing. The face area for which the higher relevance factor is calculated is used, to create mask data.

Journal ArticleDOI
TL;DR: A new method within the framework of principal component analysis (PCA) to robustly recognize faces in the presence of clutter by learning the distribution of background patterns and it is shown how this can be done for a given test image.
Abstract: We propose a new method within the framework of principal component analysis (PCA) to robustly recognize faces in the presence of clutter. The traditional eigenface recognition (EFR) method, which is based on PCA, works quite well when the input test patterns are faces. However, when confronted with the more general task of recognizing faces appearing against a background, the performance of the EFR method can be quite poor. It may miss faces completely or may wrongly associate many of the background image patterns to faces in the training set. In order to improve performance in the presence of background, we argue in favor of learning the distribution of background patterns and show how this can be done for a given test image. An eigenbackground space is constructed corresponding to the given test image and this space in conjunction with the eigenface space is used to impart robustness. A suitable classifier is derived to distinguish nonface patterns from faces. When tested on images depicting face recognition in real situations against cluttered background, the performance of the proposed method is quite good with fewer false alarms.

Proceedings ArticleDOI
15 Oct 2005
TL;DR: This paper presents the results to evaluate the effectiveness of MPEG 7 color descriptors in visual surveillance retrieval problems, and results are presented on innovative methods to combine the output from different descriptors, and also different components of the observed people.
Abstract: This paper presents the results to evaluate the effectiveness of MPEG 7 color descriptors in visual surveillance retrieval problems. A set of image sequences of pedestrians entering and leaving a room, viewed by two cameras, is used to create a test set. The problem posed is the correct identification of other sequences showing the same person as contained in an example image. Color descriptors from the MPEG7 standard are used, including dominant color, color layout, color structure and scalable color experiments are presented that compare the performance of these, and also compare automatic and manual techniques to examine the sensitivity of the retrieval rate on segmentation accuracy. In addition, results are presented on innovative methods to combine the output from different descriptors, and also different components of the observed people. The evaluation measure used is the ANMRR, a standard in content-based retrieval experiments.

Patent
18 May 2005
TL;DR: In this paper, a method of determining a location of a component on a workpiece is presented, where a before-placement standard image is acquired of an intended placement location on a standard workpiece.
Abstract: The present invention includes a method of determining a location of a component on a workpiece. A before-placement standard image is acquired of an intended placement location on a standard workpiece. Then, a standard component is placed upon the standard workpiece and the placement is verified. An after-placement standard image is acquired and a standard difference image is created from the before and after standard images. Then, a before-placement test image is acquired of an intended placement location on the workpiece. A component is then placed upon the workpiece, and after-placement test image is acquired. A test difference image is created from the before and after test images. A first offset is calculated between the before standard difference image and the before test image. Then, the test difference is transformed based on the first offset to generate a difference test image (DTR) that is registered to the standard difference image. The standard difference image is correlated to the registered difference test image (DTR) to generate a registration offset indicative of placement efficacy.

Book ChapterDOI
16 Oct 2005
TL;DR: This work uses a decision tree of Adaboost-based classifiers, able to select the most effective classifier at every stage, based on the outcomes of the classifiers already applied to detect faces in a test image.
Abstract: While there has been a great deal of research in face detection and recognition, there has been very limited work on identifying the expression on a face. Many current face detection projects use a [Viola/Jones] style “cascade” of Adaboost-based classifiers to interpret (sub)images — e.g. to identify which regions contain faces. We extend this method by learning a decision tree of such classifiers (dtc): While standard cascade classification methods will apply the same sequence of classifiers to each image, our dtc is able to select the most effective classifier at every stage, based on the outcomes of the classifiers already applied. We use dtc not only to detect faces in a test image, but to identify the expression on each face.