48 results about "Local statistics" patented technology

A technique is provided for automatically generating a bone mask in CTA angiography. In accordance with the technique, an image data set may be pre-processed to accomplish a variety of function, such as removal of image data associated with the table, partitioning the volume into regionally consistent sub-volumes, computing structures edges based on gradients, and / or calculating seed points for subsequent region growing. The pre-processed data may then be automatically segmented for bone and vascular structure. The automatic vascular segmentation may be accomplished using constrained region growing in which the constraints are dynamically updated based upon local statistics of the image data. The vascular structure may be subtracted from the bone structure to generate a bone mask. The bone mask may in turn be subtracted from the image data set to generate a bone-free CTA volume for reconstruction of volume renderings.

A video watermarking scheme is disclosed, which is designed for the digital cinema format, as it will be used on large projector screens in theaters. The watermark is designed in such a way that it has minimal impact on the video quality, but is still detectable after capture with a handheld camera and conversion to, for instance, VHS, CD-Video or DVD format. The proposed watermarking system only exploits the temporal axis. This makes it invulnerable to geometrical distortions generally caused by such a way of capturing. The watermark is embedded by modulating a global property of the frames (e.g. the mean luminance) in accordance with the samples of the watermark. The embedding depth is preferably locally adapted within each frame to local statistics of the respective image. Watermark detection is performed by correlating the watermark sequence with extracted mean luminance values of a sequence of frames.

A global and local statistics controlled noise reduction system in which the video image noise reduction processing is effectively adaptive to both image local structure and global noise level. A noise estimation method provides reliable global noise statistics to the noise reduction system. The noise reduction system dynamically / adaptively configures a local filter for processing each image pixel, and processes the pixel with that local filter. The filtering process of the noise reduction system is controlled by both global and local image statistics that are also computed by the system.

A digital media codec adaptively re-arranges a coefficient scan order of transform coefficients in accordance with the local statistics of the digital media, so that the coefficients can be encoded more efficiently using entropy encoding. The adaptive scan ordering is applied causally at encoding and decoding to avoid explicitly signaling the scan order to the decoder in the compressed digital media stream. For computational efficiency, the adaptive scan order re-arranges the scan order by applying a conditional exchange operation on adjacently ordered coefficient locations via a single traversal of the scan order per update of the statistical analysis.

A noise reduction system that not only preserves details in images but also provides essentially clean, smooth, and natural looking homogeneous regions in images. The noise reduction system utilizes a dual-channel adaptive noise reduction technique. The input signal is first split into two channels (i.e., a low-pass channel and a high-pass channel), by a channel splitting filter. Then the two channel signals are processed separately. The low-pass channel signal is processed using an adaptive directional filter based on the estimation of the local 2D and 1D statistics and the detection of the local image structure direction. The high-pass channel signal is processed by a non-linear filtering method based on the estimation of the local statistics and the noise level of the high-pass channel signal, which is derived from the noise level of the original input signal. The processed signals from the two channels are summed together to get the final output.

The invention discloses a K-SVD (K-means singular value decomposition) speckle inhibiting method based on SAR (synthetic aperture radar) image local statistic characteristic, mainly solving the problem that detail information such as edge, texture and the like is fuzzy in the traditional speckle inhibiting method. The method is realized in the following processes of: inputting an SAR image, extracting overlapped blocks in the SAR image to obtain an overlapped block vector set; then randomly sampling the overlapped block vector set to obtain a training sample set; carrying out SAR_KSVD dictionary training on a training sample to obtain a final training dictionary; carrying out SAR_OMP sparse coding on the overlapped block vector set under the condition of the final training dictionary to obtain a sparse coding coefficient; and obtaining a speckle inhibited image by utilizing the final training dictionary and the sparse coding coefficient according to the redundant sparse representation image noise inhibiting theory. By applying the method disclosed by the invention, speckle noise in a homogenous region can be effectively inhibited, brightness and edge texture of a target at a strong reflection point can be well maintained to be clear, and the method disclosed by the invention can be applicable to SAR images in the fields such as land resource monitoring, natural disaster analysis and the like.

A technique is provided for segmenting a structure of interest from a volume dataset. The technique identifies regions of the structure using templates having characteristics of the structure of interest. The identified regions may then undergo a constrained growth process using dynamic constraints that may vary based on local statistics associated with the identified structure regions. Edges within the volume may be determined using gradient data determined by evaluating the strongest gradient between each pixel and all adjacent pixels. The edge data may be used to prevent the constrained growing process from exceeding the boundaries of the structure of interest.

Methods and systems for detecting defects on a wafer using adaptive local thresholding and color filtering are provided. One method includes determining local statistics of pixels in output for a. wafer generated using an inspection system, determining which of the pixels are outliers based on the local statistics, and comparing the outliers to the pixels surrounding the outliers to identify the outliers that do not belong to a cluster of outliers as defect candidates. The method also includes determining a value for a difference in color between the pixels of the defect candidates and the pixels surrounding the defect candidates. The method further includes identifying the defect candidates that have a value for the difference in color greater than or equal to a predetermined value as nuisance defects and the defect candidates that have a value for the difference in color less than the predetermined value as real defects.

Disclosed is a method and system to learning mid-level features for text images that leverages character bounding box annotations. According to an exemplary embodiment, the disclosed method and system includes extracting semantic local descriptors by aggregating local statistics of small patches and correlating them with character bounding box annotations.

The invention discloses a motion-information-base spatial moving object detection method of a video image. Denoising processing is carried out on a given original image by mean filtering; a single frame of image is separated by using a variable threshold value based on local statistics; an image point coordinate is calculated by using a gray weighted centroid method; and then an object is detected based on differences between imaging motion features of a fixed star and an object during satellite staring observation. The method has the following advantages: firstly, because the variable threshold value based on local statistics is used for separating the single frame of image, a phenomenon that the background intensity distribution of the spaced-based shot image is not uniform due to stray light is fully considered, so that a problem that an image can not be separated by a global threshold value can be solved; and secondly, an object is detected only based on the motion information without any fixed start rectification process, thereby substantially accelerating the object detection speed.

A method, apparatus, and article of manufacture for deriving and using non-linearly adapted lapped transforms (NALTs) In one embodiment, the method comprises receiving parameters of a statistical fit between local statistics of original and processed frame data, determining adaptation parameters for application of an inverse lapped transform to frame data, and filtering the frame data in response to filtering adaptation parameters.

The invention discloses a test system and method. The system comprises a control device and multiple test devices connected with the control device, wherein the control device is used for sending a test task to all the test devices, and all the test devices are used for testing a tested object according to the received test task; each test device is further used for performing statistical analysis on test data relevant to the test device, obtaining a local statistical result corresponding to the test device and supplying the local statistic result to the control device; and the control device is further used for summarizing the local statistical results supplied by all the test devices and determining test result data of the tested object according to the summarizing result. Through the test system and method, the processing efficiency of the control device is improved, the system resource occupation quantity of the control device is lowered, the quantity of system resources capable of being shared by all the test devices is not influenced, therefore, higher pressure can be exerted according to the tested object, and the property limit of the tested object can be accurately tested.

The invention relates to a digital image splicing passive detection method based on a frequency domain local statistic model. The method comprises steps of firstly extracting integrated frequency domain LTP characteristics of each image and performing dimension reduction processing for standardized characteristics via kernel principal component analysis (KPCA); secondly, inputting data of the characteristics going through the dimension reduction processing into an SVM classifier for training so as to obtain a trained classifier; and at last, performing standardization and dimension reduction processing for the frequency domain LTP characteristics of to-be-measured images according to the above method and performing identification for authenticity of the to-be-measured images by use of the trained SVM classifier. The method is remarkably advantaged by detection precision and calculation efficiency.

The invention discloses a real-time parallel classification method for mass data flow. The method includes the following steps of 1, obtaining of data Spout; 2, filtering and batching of Bolt; 3, obtaining of the model Bolt; 4, local statistics and calculation of Bolt; 5, evaluation of the Bolt. According to 3V characteristics including Volume, Velocity and Value in '4V' characteristics of big data and the requirement for efficient treatment of mass data, a vertical and parallel P-VFDT algorithm based on a Storm platform is achieved; it is proved by experiments on large-scale data that the P-VFDT algorithm and a VFDT algorithm have similar classification performance, time consumption of the P-VFDT algorithm in a single-computer multi-core environment is reduced by 12% compared with the VFDT algorithm, and the P-VFDT algorithm in a clustering environment is reduced by about 8% compared with the VFDT algorithm.

The invention discloses a radar networking airspace target detection method based on local statistic fusion. The implementation of the radar networking airspace target detection method comprises the steps of receiving an echo signal, and sampling the echo signal; estimating the noise power of a signal to be detected in the echo signal; constructing local detection statistics, and determining the detection performance of monostatic radar; carrying out a summation adding operation on the local detection statistics to obtain a global detection statistic, and determining the detection performanceof radar networking; and determining whether there is a target or not in the echo signal by using the detection performance of the radar networking. According to the invention, the noise power of themonostatic radar is estimated in the detection process, there is no need to estimate the detection threshold when the existence of a target is determined at the fusion center, the target is consideredto exist if the global detection statistic is greater than 0, and the detection speed of the radar networking system for the target is improved.

Methods and systems for detecting defects on a wafer using adaptive local thresholding and color filtering are provided. One method includes determining local statistics of pixels in output for a wafer generated using an inspection system, determining which of the pixels are outliers based on the local statistics, and comparing the outliers to the pixels surrounding the outliers to identify the outliers that do not belong to a cluster of outliers as defect candidates, The method also includes determining a value for a difference in color between the pixels of the defect candidates and the pixels surrounding the defect candidates. The method further includes identifying the defect candidates that have a value for the difference in color greater than or equal to a predetermined value as nuisance defects and the defect candidates that have a value for the difference in color less than the predetermined value as real defects,

The invention discloses a multi-scale-based local statistic active contour model (LSACM) level set image segmentation method. The offset field B epsilon, a variance sigma i epsilon and a level set function Phi(x) in an LSACM level set method are initialized. The quantity L(x) for describing a local area characteristic in a multi-scale LSACM method is calculated. A differential characteristic d(X) which describes a multi-scale local area is calculated. The maximum response M of a high pass filter in the multi-scale LSACM method is calculated. A local area simulation gray Ci epsilon is updated. The offset field B epsilon is updated. The variance sigma i epsilon is updated. The purpose of curve evolution is achieved through solving the partial differential equation minimum value corresponding to a multi-scale LSACM level set energy function. If a set number of iterations is achieved, the iteration operation is stopped, the curve evolution is ended, and if the number of iterations is not achieved, the iteration is continued. The invention provides the multi-scale LSACM level set method, a gray uneven image can be effectively segmented, and the phenomena of excessive segmentation and insufficient segmentation in an image segmentation method are improved.

A method, apparatus, and article of manufacture for deriving and using non-linearly adapted lapped transforms (NALTs) In one embodiment, the method comprises receiving parameters of a statistical fit between local statistics of original and processed frame data, determining adaptation parameters for application of an inverse lapped transform to frame data, and filtering the frame data in response to filtering adaptation parameters.

The invention provides a hole site protection door state detection method and device, computer equipment and a storage medium. The method comprises steps of starting from image recognition, acquiringan image of to-be-tested equipment to construct a training sample set; obtaining a preset threshold value of a discrimination function straight line according to the training sample set; then, carrying out self-adaptive dynamic threshold segmentation based on local statistics on a newly obtained image of the to-be-detected equipment, accurately screening out the hole site area; projecting extracted gray features of the hole site area to a preset discrimination function straight line to achieve the effects of extracting the classification information and compressing the feature space dimension;wherein after projection, the feature subset has the maximum inter-class distance and the minimum intra-class distance in the subspace after projection; finally, judging whether the to-be-detected equipment has defects or not according to the projection ordinate value of the gray feature subset and the threshold value of the preset discrimination function straight line. The method is convenient,fast, effective, high in accuracy, high in anti-interference capacity and stable in detection result.

The invention discloses a method for removing mixed noise of a color image, which relates to the technical field of image processing. The invention performs denoising processing on the mixed noise composed of impulse noise and Gaussian noise based on the block matching filter and combined with the local characteristics of the image. At first, the impulse noise is denoised by utilizing the local statistic characteristic of the image, and then 3-D matching filtering, namely Gaussian denoising is performed according to the global sparsity and autocorrelation of the image. The invention fully utilizes the information characteristics of the image, so the invention can more effectively remove the mixed noise of the color image compared with the prior method, and is more suitable for the denoising processing of the natural image with high resolution and large color depth, and has strong practicability.

The invention relates to a data fusion method for a multi-sensor distributed system. According to the condition that the change of signal to noise ratios of different sensors is great, the data fusion method utilizing the local statistic power for summation is provided. The data fusion method comprises the steps that each local sensor stores N received pulse data, calculates the local R statistic according to the received data and transmits the R statistic to an fusion center; the fusion center calculates global weighing summation statistic R by utilizing the statistic of each local sensor and carries out detection based on the global weighing summation statistic.

The invention discloses a method for determining the contraction resistances of a contact interface under different pressure intensities, and the method comprises the steps: carrying out the local statistic observation of the contact interface through a measurement instrument, and obtaining the information of micro-bulges which represent the actual surface shape, wherein the information comprises micro-bulge height value distribution mode, micro-bulge height standard deviation, micro-bulge surface density and micro-bulge curvature radius; finally obtaining the contraction resistances of the contact interface under different pressure intensities through the information. The method can track and obtain the state and change information of each micro-bulge on the surface, can be suitable for any micro-bulge height distribution mode, can calculate the contraction resistance under the DC condition, and also can calculate the contraction resistance of the contact interface under the AC condition. The result can be used for the research of an equivalent circuit model of a microwave part contact interface, and provides support for the research of a passive intermodulation nonlinear mechanism of a microwave part.