Abstract: A system and method for tracking a tumor includes a regression module for selecting, using a motion signal and a regression function, a feature signal from a set of feature signals, each feature signal in the set of feature signals represents a medical image of the body of the patient, wherein the motion signal represents a motion of a surface of a skin of the patient caused by the respiration, and wherein the regression function is trained based on a set of observations of the motion signal synchronized with the set of feature signals; and a registration module for determining the location of the target object using the feature signal and a registration function, wherein the registration function registers each feature signal to a breath-hold location of the target object identified.

Abstract: A method estimates a pattern of change of a patient, specifically a change in the respiration pattern. An ultrasound video is segmented into groups of pictures (GOPs). Pixels from the first GOP are used to initialize a change model. Based on the change model, a change pattern for a next GOP is estimated, and the change model is changed to fit the change pattern. The estimating and the updating are repeated until a termination condition is reached.

Abstract: A method detects a target in a sequence of radar images, wherein each image is partitioned into a grid of cells, and wherein each cell has a corresponding position in an image coordinate system associated with a location in a world coordinate system. For each most recent image in a sliding temporal window of images, intensities of each cell are determined, and the subset of the cells having highest intensities is stored as a set of current needles. A set of hypotheses, obtained by using a state transition model and corresponding maximum limits, is determined for the current set of needles and appended to a set of queues. The hypotheses for the previous sets of needles to the corresponding set of queues are updated, and a maximum likelihood in the set of queues are selected to detect the location of targets.

Abstract: Embodiments of the invention provide a method and a system for tracking an object from a training image to a target image. The training image and the target image are elements of a sequence of images. The object in the training image is represented by an object state. First, a set of particles is acquired, wherein each particle in the set of particles is associated with a weight, such that the particle represents the object state with a probability equal to the weight. Next, a regression function is applied to each particle in the set of particles based on a target image to determine a set of moved particles and the object state is updated according to the set of moved particles, such that the object state represents the object in the target image.

Abstract: A generic classifier is adapted to detect an object in a particular scene, wherein the particular scene was unknown when the classifier was trained with generic training data. A camera acquires a video of frames of the particular scene. A model of the particular scene model is constructed using the frames in the video. The classifier is applied to the model to select negative examples, and new negative examples are added to the training data while removing another set of existing negative examples from the training data based on an uncertainty measure. Selected positive examples are also added to the training data and the classifier is retrained until a desired accuracy level is reached to obtain a scene specific classifier.

Abstract: A tumor is tracked in sequences of bi-plane images by generating a set of segmentation hypotheses using a 3D model of the tumor, a bi-plane geometry, and a previous location of the tumor as determined from the pairs of biplane images. Volume prior probabilities are constructed based on the set of hypotheses. Seed pixels are selected using the volume prior probabilities, and a bi-plane dual image graph is constructed using intensity gradients and the seed pixels to obtaining segmentation masks corresponding to tumor boundaries using the image intensities to determine a current location of the tumor.

Abstract: A system and method for tracking a tumor includes a regression module for selecting, using a motion signal and a regression function, a feature signal from a set of feature signals, each feature signal in the set of feature signals represents a medical image of the body of the patient, wherein the motion signal represents a motion of a surface of a skin of the patient caused by the respiration, and wherein the regression function is trained based on a set of observations of the motion signal synchronized with the set of feature signals; and a registration module for determining the location of the target object using the feature signal and a registration function, wherein the registration function registers each feature signal to a breath-hold location of the target object identified.

Abstract: A method normalizes a feature of an object in an image. The feature of the object is extracted from a 2D or 3D image. The feature is displaceable within a displacement zone in the object, and wherein the feature has a location within the displacement zone. An associated description of the feature is determined. Then, the feature is displaced to a best location in the displacement zone to produce a normalized feature.

Abstract: The embodiments of the invention describe a method for segmenting an image. We perform an initial segmentation of the image to produce a previous segmented region and segment iteratively the image using a spatial random walk based on a shape prior of the previous segmented region to produce a next segmented region. We compare the next segmented region with the previous segmented region, and repeat the segmenting and the comparing until the previous and next segmented regions converge. After that, we select the next segmented region as a final segmented region.

Abstract: A road surface includes lane marking that store digital information. Images of the road surface and lane markings are acquired by a camera. The digital information is decoded from the images, analyzed so that a feedback signal can be generated according to the decoded digital information.

Abstract: The current invention describes a method for filtering an input image with a bilateral filter. The bilateral filter includes a spatial filter and a range filter. The method constructs a set of power images from an input image including pixels, each pixel having intensity. Then, the method applies, to each power image, the spatial filter to determine a response for the spatial filter and the corresponding power image and combines the responses and the set of power images to produce a response for the bilateral filter.

Abstract: A method trains a multi-class classifier by iteratively performing the following steps until a termination condition is reached. The probabilities of class membership for unlabeled data obtained from an active pool of unlabeled data are estimated. A difference between a largest probability and a second largest probability is determined. The unlabeled data with the lowest difference is selected, labeled and then added to a training data set for training the classifier.

Abstract: Region of interest (ROI) corresponding to a soft tissue mass are tracked in a training video acquired by sonography. The locations of the ROI are used to construct a directed graph in which each node represents a location of the tracked ROI, and the edges represent temporal relations of the ROIs. The soft tissue mass can also be tracked using the graph, and appropriate treatment can be delivered.

Abstract: The current invention describes a method for filtering an input image with a bilateral filter to produce an output image. The bilateral filter includes a spatial filter and a range filter.

Abstract: A classifier for detecting objects in images is constructed from a set of training images. For each training image, features are extracted from a window in the training image, wherein the window contains the object, and then randomly sample coefficients c of the features. N-combinations for each possible set of the coefficients are determined. For each possible combination of the coefficients, a Boolean valued proposition is determined using relational operators to generate a propositional space. Complex hypotheses of a classifier are defined by applying combinatorial functions of the Boolean operators to the propositional space to construct all possible logical propositions in the propositional space. Then, the complex hypotheses of the classifier can be applied to features in a test image to detect whether the test image contains the object.

Abstract: A generic classifier is adapted to detect an object in a particular scene, wherein the particular scene was unknown when the classifier was trained with generic training data. A camera acquires a video of frames of the particular scene. A model of the particular scene model is constructed using the frames in the video. The classifier is applied to the model to select negative examples, and new negative examples are added to the training data while removing another set of existing negative examples from the training data based on an uncertainty measure;. Selected positive examples are also added to the training data and the classifier is retrained until a desired accuracy level is reached to obtain a scene specific classifier.

Abstract: A method recognizes a set of traffic signs in a sequence of images acquired of a vehicle environment by a camera mounted in a moving vehicle by detecting in each image, a region of interest (ROI) using a parameter space transform. The ROI is tracked and classified as a particular one of the signs. The classifier only uses a same class and a different class, and a regression function to update the classifier.

Abstract: A method detects a target in a sequence of radar images, wherein each image is partitioned into a grid of cells, and wherein each cell has a corresponding position in an image coordinate system associated with a location in a world coordinate system. For each most recent image in a sliding temporal window of images, intensities of each cell are determined, and the subset of the cells having highest intensities is stored as a set of current needles. A set of hypotheses, obtained by using a state transition model and corresponding maximum limits, is determined for the current set of needles and appended to a set of queues. The hypotheses for the previous sets of needles to the corresponding set of queues are updated, and a maximum likelihood in the set of queues are selected to detect the location of targets.

Abstract: Invention describes a method and system for detecting and tracking an object in a sequence of images. For each image the invention determines an object descriptor from a tracking region in a current image in a sequence of images, in which the tracking region corresponds to a location of an object in a previous image. A regression function is applied to the descriptor to determine a motion of the object from the previous image to the current image, in which the motion has a matrix Lie group structure. The location of the tracking region is updated using the motion of the object.

Abstract: A method recognizes a set of traffic signs in a sequence of images acquired of a vehicle environment by a camera mounted in a moving vehicle by detecting in each image, a region of interest (ROI) using a parameter space transform. The ROI is tracked and classified as a particular one of the signs. The classifier only uses a same class and a different class, and a regression function to update the classifier.