Showing papers on "Standard test image published in 1999"

Content-based hierarchical classification of vacation images

Abstract: Grouping images into (semantically) meaningful categories using low level visual features is a challenging and important problem in content based image retrieval. Using binary Bayesian classifiers, we attempt to capture high level concepts from low level image features under the constraint that the test image does belong to one of the classes of interest. Specifically, we consider the hierarchical classification of vacation images; at the highest level, images are classified into indoor/outdoor classes, outdoor images are further classified into city/landscape classes, and finally, a subset of landscape images is classified into sunset, forest, and mountain classes. We demonstrate that a small codebook (the optimal size of codebook is selected using a modified MDL criterion) extracted from a vector quantizer can be used to estimate the class-conditional densities of the observed features needed for the Bayesian methodology. On a database of 6931 vacation photographs, our system achieved an accuracy of 90.5% for indoor vs. outdoor classification, 95.3% for city vs. landscape classification, 96.6% for sunset vs. forest and mountain classification, and 95.5% for forest vs. mountain classification. We further develop a learning paradigm to incrementally train the classifiers as additional training samples become available and also show preliminary results for feature size reduction using clustering techniques.

Feature-based defect detection

Abstract: Methods and apparatus are provided for inspecting a patterned substrate, comprising: preparing a reference image and a test image, extracting features from the reference image and extracting features from the test image, matching features of the reference image and features of the test image; and comparing features of the reference image and of the test image to identify defects. Embodiments include apparatus for inspecting patterned substrates, computer-readable media containing instructions for controlling a system having a processor for inspecting patterned substrates, and computer program products comprising a computer usable media having computer-readable program code embodied therein for controlling a system for inspecting patterned substrates. The images can be electron-beam voltage-contrast images.

Super-resolution of images based on local correlations

Abstract: An adaptive two-step paradigm for the super-resolution of optical images is developed in this paper. The procedure locally projects image samples onto a family of kernels that are learned from image data. First, an unsupervised feature extraction is performed on local neighborhood information from a training image. These features are then used to cluster the neighborhoods into disjoint sets for which an optimal mapping relating homologous neighborhoods across scales can be learned in a supervised manner. A super-resolved image is obtained through the convolution of a low-resolution test image with the established family of kernels. Results demonstrate the effectiveness of the approach.

Image Stitching - Comparisons and New Techniques

Abstract: This work describes the steps involved in the generation of a panoramic image, i.e., image acquisition, image registration and image merging. Different approaches for each of these steps are discussed and compared. The resultant images from image registration and image merging are quantitatively evaluated to provide an indication to the performance of the different methods. The methods provided in this work can be used to generate panoramic images for the use of interactive panoramic viewing of images, architectural walk-through, and other applications associated with the modelling of 3D environments.

Image projecting device

Abstract: PROBLEM TO BE SOLVED: To save the labor of an installation adjustment work of a device for projecting an image on a screen. SOLUTION: In the state where an image projected by a projector 2 arranged obliquely relative to a screen 1 having a free curved surface is observed at a certain viewpoint position 7, a test image is projected, and the test image is photographed by a camera 4 on the viewpoint position 7, and correction data for giving reverse strain beforehand is produced beforehand, and correction treatment of the image required to be projected is executed by the correction data, and the image is projected by the projector 2, and hereby a right image having no strain can be obtained, viewing from the viewpoint position 7.

Technique for watermarking an image and a resulting watermarked image

Abstract: Apparatus and an accompanying method, for generating a cryptographic identifier for a non-marked image and embedding that identifier within the image itself in order to generate a “watermarked” image; for subsequently detecting that watermark in a test image; and the watermarked image so formed. First, pixel values for a non-marked image are transformed, either directly or after being enhanced, into a series of transform coefficients. A set of pseudo-random perturbation values which collectively constitute the watermark is determined wherein each of these values is heuristically selected, such that all these values collectively satisfy a plurality of different mathematical constraints and each such value preferably equals a relatively small value in a predefined range. These perturbation values are then added to the transform coefficients. Resulting perturbed coefficients are then inversely transformed back to pixel values to form the watermarked image. To detect whether a test image contains the watermark and hence is a copy of the watermarked image, the pixel values for the test image are transformed to yield transform coefficients. A plurality of different mathematical tests or a majority type rule is then used, in conjunction with the perturbation values, previously used to create the watermark, and these transform coefficients to determine whether the perturbation values collectively exist in the test image, and hence whether the watermark is present or not.

Higher order statistical learning for vehicle detection in images

Abstract: The paper describes a scheme for detecting vehicles in images. The proposed method approximately models the unknown distribution of the images of vehicles by learning higher order statistics (HOS) information of the 'vehicle class' from sample images. Given a test image, statistical information about the background is learnt 'on the fly'. An HOS-based decision measure then classifies test patterns as vehicles or otherwise. When tested on real images of aerial views of vehicular activity, the method gives good results even on complicated scenes. It does not require any a priori information about the site. However, it is amenable to augmentation with contextual information. The method can serve as an important step towards building an automated roadway monitoring system.

Digipaper: a versatile color document image representation

Abstract: We describe a segmentation method and associated file format for storing images of color documents. We separate each page of the document into three layers, containing the background (usually one or more photographic images), the text, and the color of the text. Each of these layers has different properties, making it desirable to use different compression methods to represent the three layers. The background layers are compressed using any method designed for photographic images, the text layers are compressed using a token-based representation, and the text color layers are compressed by augmenting the representation used for the text layers. We also describe an algorithm for segmenting images into these three layers. This representation and algorithm can produce very highly-compressed document files that nonetheless retain excellent image quality.

Method and apparatus for automated image quality evaluation of X-ray systems using any of multiple phantoms

Abstract: A system for automated x-ray system parameter evaluation is provided. A physical model or template is created and stored in the system, one for each desired phantom. The automated system imports a grayscale x-ray image and then processes the image to determine image components. First, a histogram of the image is created, then a threshold in the histogram is determined and the imported image is binarized with respect to the threshold. Next, connected component analysis is used to determine image components. If the components do not match, then the image is rejected. The system next locates landmarks in the imported image corresponding to expected physical structures. The landmarks include a perimeter ring, vertical and horizontal line segments, and fiducials. The system uses the landmarks to predict Regions of Interest (ROIs) where measurement of the x-ray system parameters takes place. Finally, the x-ray system parameters are measured in the identified ROIs.

Technique for detecting a watermark in a marked image

Abstract: Apparatus and an accompanying method, for generating a cryptographic identifier for a non-marked image and embedding that identifier within the image itself in order to generate a “watermarked” image; for subsequently detecting that watermark in a test image; and the watermarked image so formed. First, pixel values for a non-marked image are transformed, either directly or after being enhanced, into a series of transform coefficients. A set of pseudo-random perturbation values which collectively constitute the watermark is determined wherein each of these values is heuristically selected, such that all these values collectively satisfy a plurality of different mathematical constraints and each such value preferably equals a relatively small value in a predefined range. These perturbation values are then added to the transform coefficients. Resulting perturbed coefficients are then inversely transformed back to pixel values to form the watermarked image. To detect whether a test image contains the watermark and hence is a copy of the watermarked image, the pixel values for the test image are transformed to yield transform coefficients. A plurality of different mathematical tests or a majority type rule is then used, in conjunction with the perturbation values, previously used to create the watermark, and these transform coefficients to determine whether the perturbation values collectively exist in the test image, and hence whether the watermark is present or not.

SamMatch: a flexible and efficient sampling-based image retrieval technique for large image databases

Abstract: The rapid growth of digital image data increases the need for efficient and effective image retrieval systems. Such systems should provide functionality that tailors to the user's need at the query time. In this paper, we propose a new image retrieval technique that allows users to control the relevantness of the results. For each image, the color contents of its regions are captured and used to compute similarity. Various factors, assigned automatically or by the user, allow high recall and precision to be obtained. We implemented the proposed technique for a large database of 16,000 images. Our experimental results show that this technique is not only space-time efficient but also more effective than recently proposed color histogram techniques.

Quantitative analysis system and method for certifying ultrasound medical imaging equipment

Abstract: An analysis system and method provide for quantitatively evaluating image quality characteristics of an ultrasound imaging machine that evaluates at least one image representation of a standard phantom acquired by the image machine. The machine under test by comparing acquired parameters with prestored values, and returning a determined set of image quality indices, along with a single index representing an arithmetic combination of all other image quality indices, which indicate the accuracy of the test image relative to a “gold standard” that has been pre-established for the model of imaging machine under investigation. The system, which includes a computer-programmed set of instructions and data, optionally includes at least one standard phantom. The image quality indices, or metrics, quantitatively represent an evaluation of a test image using a set of relatively subjective criteria that include homogeneity, contrast, signal attenuation and penetration of depth, pin to background ratio in near and far-field, axial and lateral resolution, modulation transfer function, and geometric distortion, and axial and lateral linearity. These image quality indices are determined by specific algorithms and then combined to form an image health index. The image health index and the individual component indices are compared to a gold standard set of indices obtained from an equivalent imaging machine operating under optimum conditions and settings.

Real-time image mosaicing from a video sequence

Abstract: The paper describes a fast and robust image registration method that can be used to create a panoramic image/video from video sequences. To estimate alignment parameters for image registration, the method computes pseudo motion vectors that are rough estimates of optical flows at each selected pixel. Using the proposed method, we implemented a software system that can, with a low-cost PC, create and display panoramic images/videos in real time.

Compression of compound documents

Abstract: Compound (or mixed) document images contain graphic or textual content along with pictures. They are a very common form of documents, found in magazines, brochures, web-sites etc. Because of the very distinct nature of those two image classes (text/graphics vs. pictures), their compression invariably involves multiple compression systems and a region segmentation (classification) method. We review state-of-the-art technologies on the subject while focusing our attention on the mixed raster content (MRC) multi-layer approach. We also present new results on segmentation for MRC based on optimized rate-distortion-based block thresholding.

Method to allow automated image quality analysis of arbitrary test patterns

Abstract: An image quality analysis system is provided that can determine various aspects of image quality easily and with minimal user involvement and minimal user expertise. The system uses a scanner, either a stand-alone or part of a multi-function printer/scanner/copier, to scan a printed test pattern, and then perform a series of analyses on the scanned image using an image quality analysis module that may be built into the image output device being tested, or provided as a stand-alone component that can receive the output from the scanner. There are often a number of different test patterns that would be used depending on which print quality issues are being tested. By encoding each test pattern with a coded identification label, not only can the particular test pattern be identified, but the analysis to be performed can also be determined from the corresponding script when the printed test pattern is scanned and subsequently decoded. Particularly suitable decoders are OCR and barcode readers. Such identification labels and/or scripts can also be used to identify a particular machine that is being tested and other relevant information.

Method and system for detecting errors in a sample image

Abstract: A method and system are described for use at a machine vision station to correctly identify errors within a sample image of an object by means of a probability image where a value of each pixel of the probability image is a probability. The probability image is created from a mean template image and a standard deviation image (46), both of which are created from a plurality of reference images of the object (40). The probability image is processed to detect the errors in the sample image independent of the values of the pixels which make up the same image. Two different techniques are described for creating the probability image utilizing mean template, sample and standard deviation pixel values.

Image processing apparatus and method for processing a color image, and storage medium storing program code of the method

Abstract: An image processing apparatus which obtains an output result faithful to an original image independently of characteristics of a device to optically read the original image. When the original image is read by an image scanner, layout and the like of the read image are analyzed, recognition is performed on characters, and compression is performed on an image area. The image data is stored with Profile (color characteristic information) unique to the image scanner. When the image is displayed or print-outputted, color matching is performed in accordance with Profile of a display device or printer and the Profile of the image scanner, and the image is reproduced.

Authenticity and integrity of digital mammographic images

Abstract: Image authenticity and integrity is an important issue in a telemammography system. We present an algorithm which can embed encrypted image and patient information into an image. The embedded information can be extracted and decrypted by the receiving site to verify the patient identification and confirm image authenticity and integrity. Because of the large size of mammographic images comparing to other images, data embedding in a mammogram is relatively easier to be implemented. By analyzing the noise gray level of our digital mammography system, we know that the least-significant bits of the image are noise caused by the imaging device. So these bits can be used for data embedding. The methods include: (I) Calculate the check-sum of the image and extract patient information from DICOM header, (2) Encrypt the check-sum and patient information using public-key encryption strategy, (3) Generate a set of uniformly distributed pseudo-random numbers and put the encrypted check-sum and patient information into randomly selected pixel locations. Three mammographic images are selected for our experiment. Two images are the digitized mammogram with a large breast and a small breast, and the other is the direct digital mammogram. About 500 characters of patient information and a 32 bits check-sum are embedded into each image. By comparing the original image and the embedded image from a 2k x 2k monitor, we found three embedded images of large size and small size of digitized mammogram as well as direct digital mammogram have no quality degradation. To prove the effectiveness of this method, we change the value of one pixel which is selected randomly in the embedded image. Then, we use our extraction algorithm to detect the integrity of this image. The results show we can not only extract the embedded patient information correctly, but also detect the slight difference between the original image and the altered image. Our preliminary results demonstrate that embedding extra information into an image using data hiding technology is an effective method for image integrity in telemammography.

Medical image data compression employing image descriptive information for optimal compression

Abstract: An image data compression technique is disclosed in which image data compression is based at least partially upon information encoded in a descriptive portion of an image data file. The descriptive portion may be a header appended to image data. In medical diagnostic applications the descriptive data may include DICOM compliant data specifying imaging modality, anatomy or other features viewable in the reconstructed images, image matrix size, and so forth. Based upon the descriptive data, the image data files are analyzed and optimal compression algorithms are identified and selected. The image data files are then compressed in accordance with the selected algorithms. The algorithms may include appropriate compression code tables, desired lengths of subregions into which the image data stream is divided for compression, predictor algorithms used to determine relative entropy levels within the image data, and so forth.

Method for assessing overall quality of digital images

Abstract: A method for producing a numerical representation of perceived overall image quality of a digital test image, including creating a digital reference image series of a reference scene which differ from each other in at least one perceptual attribute, wherein each digital reference image has a corresponding numerical representation of its overall image quality; iteratively displaying one of the digital reference images and comparing that digital reference image with the displayed digital test image and continuing the process until the numerical representation of the digital test image may be satisfactorily inferred; and recording the numerical representation corresponding to the digital test image.

Method for optimizing compression of scanned data

Abstract: A method for choosing a compression process for a scanned image includes scanning an image, forming a bitmap of the scanned image and selecting at least one portion of the scanned image as a compression test subject. The selected portion is selected by a variety of compression techniques and a score for each technique is assigned. The compression technique with the best score is then used to compress the entire bitmap of the scanned image.

Method and system for capturing large format documents using a portable hand-held scanner

Abstract: A scanning device includes a mode for generating a reproduced image on a standard page size from a large-sized original image. In this mode, the sampling rate of the scanning device in a motion direction is reduced. In addition, the captured image data in the sensor direction is downsampled by some factor. Position data that is generated as the image data is captured is also modified in accordance with the downsampling factor. A reproduced image is then generated from the modified image and position data and printed at a standard resolution.

Adaptive mask technique for defect inspection

Abstract: A two-dimensional scatter plot is created by plotting the gray levels of pixels from a test image against the gray levels of corresponding pixels from a reference image. A noise reduction filter is applied on the scatter plot to define a mask shape which can be extracted and filled-in to generate a mask. Defect pixels on the test image are identified by comparing corresponding pixel gray values against the mask. A typical application is detecting defects in a semiconductor wafer during device fabrication.

Image data processing method, image data print apparatus, record medium recording image data processing program

Abstract: The characteristics of processing operation tendency, performance, etc., peculiar to image data acquisition means (digital still camera) are previously examined and the image data processing contents considering the characteristics of the digital still camera are preset in an image data processing content storage section 13 in correspondence with the model name of the digital still camera. The image data acquired with one digital still camera is read into an image data read section 11 , a model determination section 12 determines the model name, an image data processing section 14 selects the image data processing contents corresponding to the determination result in response thereto and processes the image data in accordance with the image data processing contents, and a print processing section 15 performs print processing. The image data processing to be performed includes image data correction processing for making a color correction, etc., and resizing processing for enlarging or reducing the image data corresponding to the model.

Image processing method and system

Abstract: An image processing method for image-processing digital image data acquired by a digital camera to enable reproduction of high-quality images includes the steps of computing first characteristic values characterizing colors of an image represented by digital image data, normalizing the digital image data based on the first characteristic values to produce normalized image data, computing from the normalized image data a second characteristic value characterizing density of the image represented by the digital image data, and correcting the normalized image data based on the second characteristic value to produce processed image data. Also provided is an image processing system for carrying out the method.

Image database querying using a multi-scale localized color representation

Abstract: A computationally efficient method to encode the spatial concentration of colors in images based on local histogram intersection is introduced. The representation allows image matching at multiple resolutions and partial matching of image regions without requiring image segmentation or cluster approximations. The method has been tested on a heterogeneous database of 16643 images and video frames. Results of objective evaluation using manually selected ground-truth image queries as well as ground-truth video keyframe queries are presented.

Image formation device, control method therefor and storage medium

Abstract: PROBLEM TO BE SOLVED: To suppress the generation of density irregularities in the main scanning direction of an image carrier in an image formation part at a low cost and to form excellent images. SOLUTION: The test image having prescribed density is formed and it is read by a reader part 18. In the read, the density data of the image in the main scanning direction (direction vertical to the carrying direction of a recording medium) of a photosensitive drum 1 in the image formation part are obtained. As a result, since the density characteristics of the main scanning direction in the image formation part are obtained, correction data based on them are prepared and stored in a correction table inside a main scanning irregularity correction circuit 50. Thereafter, normally, at the time of forming the read images in the main scanning direction at the time of a copying operation, pixel data there are corrected by the correction data and the images are formed.

Evaluation of an optical correlator automatic target recognition system for acquisition and tracking in densely cluttered natural scenes

Abstract: Paul C. MillerMichael RoycePeter VirgoMerrilyn FiebigGarry HamlynSurveillance Systems DivisionDefence Science and TechnologyOrganisation, SalisburyP.O. Box 1500Salisbury, South Australia 5108AustraliaE-mail: paul.miller@dsto.defence.gov.auAbstract. We describe a real-time automatic target recognition (ATR)system that employs a state-of-the-art optical correlator. The system’sability to acquire and track a moving vehicle target against a denselycluttered natural background is investigated using six test image se-quences obtained from a scanning IR sensor array mounted on an air-borne platform. The approach adopted for the evaluation is to first opti-mize a number of system parameters, using what we considered to beone of the better test sequences, and then to test the system with theoptimum configuration on the remaining sequences. The system perfor-mance is quantified by measuring its receiver-operating-characteristiccurves against each test image sequence. Although successful acquisi-tion and tracking of the target is demonstrated for some test sequences,there are numerous occasions on which the system acquired andtracked false alarms. This is primarily because edge features alone arean insufficient discriminator. We therefore conclude that the systemevaluated does not exhibit the desired robustness for the acquisition andtracking of vehicles in densely cluttered natural scenes. Finally, theseresults are compared with those produced using a digital simulation ofthe ATR system’s algorithmic processing chain. Comparison shows thatfor all practical purposes the performances are equivalent.

Focused JPEG encoding based upon automatic preidentified regions of interest

Abstract: We have developed a focused-procedure based upon a collection of image processing algorithms that serve to identify regions-of-interest (ROIs), over a digital image. To loci of these ROIs are quantitatively compared with ROIs identified by human eye fixations or glimpses while subjects were looking at the same digital images. The focused- procedure is applied to adjust and adapt the compression ratio over a digital image: - high resolution and poor compression for ROIs; low resolution and strong compression for the major expanse of the entire image. In this way, an overall high compression ratio can be achieved, while at the same time preserving, important visual information within particularly relevant regions of the image. We have bundled the focused-procedures with JPEG, so that the JPEG version allows the result of the compression to be formatted into a file compatible for standard JPEG decoding. Thus, once the image has been compressed, it can be read without difficulty.