31 results about "Scale-space segmentation" patented technology

A system and method for automatically generating a hierarchical table of contents or outline for indexing a document and identifying clusters of related information in the document. The document may comprise text, audio, video, or a multimedia presentation. The invention employs a unique and novel combination of latent semantic indexing techniques to identify related blocks and major topic changes within the document with scale space segmentation techniques to respectively identify self-similar blocks within the document and to thus find topic changes of various sizes at block edges. The invention then produces a visual presentation of the semantic structure of the document.

A scale-Space feature extraction technique is based on recursive decomposition of polyhedral surfaces into surface patches. The experimental results show that this technique can be used to perform matching based on local model structure. Scale-space techniques can be parameterized to generate decompositions that correspond to manufacturing, assembly or surface features relevant to mechanical design. One application of these techniques is to support matching and content-based retrieval of solid models. Scale-space technique can extract features that are invariant with respect to the global structure of the model as well as small perturbations that 3D laser scanning may introduce. A new distance function defined on triangles instead of points is introduced. This technique offers a new way to control the feature decomposition process, which results in extraction of features that are more meaningful from an engineering viewpoint. The technique is computationally practical for use in indexing large models.

The invention provides a high-performance implementation method for multi-scale segmentation of remote sensing images, and in particular a method for implementing fast and multi-scale image segmentation of large amount of remote sensing images and establishing the structural relationship of segmentation results in the process of high-resolution remote sensing image information extraction. The method comprises the following steps: on the basis of analyzing the implementation procedure of the algorithm and finding out the computing-intensive segment of the algorithm, realizing the parallel segmentation of the algorithm-intensive segment based on MPI and OMP models, and carrying out date join on the parallel segmentation results; implementing the multi-scale image segmentation by storing the initial segmentation results of the algorithm and merging the subsequent multi-scale remote sensing images; and establishing a multi-scale object topological relation model. The generated multi-scale segmentation region and corresponding hierarchical relationship can be applied to various services; and the corresponding implementation method is applicable to various segmentation algorithms such as mean shift and the like, and can greatly improve the data amount processed by algorithm and the processing efficiency.

The invention brings forward a Chinese word segmentation method by using character embedding based on word context and a neural network. The character embedding is learnt on large-scale automatic segmentation of data. The learnt character embedding is used as inputting of a segmentation model of the neural network so as to effectively help model learning. The method comprises the concrete steps of learning character embedding on large-scale automatic segmentation of data according to the word context and lexeme marking; and utilizing character-embedding as inputting of the segmentation model of the neural network so that segmentation performance is improved. Compared with other Chinese word segmentation method technologies based on the neural network, the method adopts the character-embedding technology based on the word context so that word information is effectively integrated into the segmentation model. Therefore, the accuracy of a word segmentation task is improved.

The invention discloses a registration method for visible light and infrared images based on multi-scale segmentation and SIFT, which comprises the following steps: step 1, preprocessing is performed to visible light images and infrared images, wherein Gaussian filtering processing is performed to the visible light images, and linear enhancement processing is performed to the infrared images; step 2, an initial parameter value which includes a scale factor, a compact factor and a shape factor is set, and the multi-scale image segmentation is performed to the visual light and infrared images subjected to preprocessing; step 3, the SIFT registration is performed to the visible light and infrared images subjected to the multi-scale image segmentation, calculation parameters of homonymy point selection transformation models are found out and the registration is performed to source images; step 4 and step 5, judgment is performed, if the visible light images and the infrared images are not subjected to accurate registration or the registration accuracy does not meet the requirement of the set threshold, the operation is returned to step 2 to adjust the parameter value of the multi-scale image segmentation method till the registration accuracy of the visual light images and infrared images meets the requirement of the set threshold, iteration is stopped, and the registration result is output.

The invention provides a high-resolution remote sensing image change detection method based on multi-scale segmentation and fusion, belongs to the technical field of hyperspectral remote sensing images, and solves the problems that the present remote sensing image change detection technology has low detection accuracy on the high-resolution remote sensing image and cannot guarantee the integrity of the detection result. The concrete process of the method comprises the steps that spatial scale segmentation is performed on the multi-temporal high-resolution remote sensing image by using a multi-scale segmentation algorithm; feature extraction is performed on the target in each scale of image after segmentation on the object perspective, and the object is described by using the object features so that vector analysis is performed relative to the remote sensing image of other temporal and object difference images of multiple scales are obtained; and change information extraction and fusion are performed on the obtained object difference images of multiple scales so that the final total change result image is obtained. The high-resolution remote sensing image change detection method is used for high-resolution remote sensing image change detection.

The invention discloses an image multi-scale segmentation method integrated with edge information. The method comprises the following steps: acquiring monochromatic, color or multiband image data; acquiring edge intensity data of the above image data; setting one or more image layer of the above image data as an image set to be segmented; setting an edge constrained condition, a segmentation method and a scale increase mode, and segmenting an image with a continuously increasing scale coefficient; carrying out image multi-scale segmentation under a scale coefficient, before merging of any patch, calculating statistics edge intensity in the patch after merging, and when the statistics edge intensity is larger than the edge constrained condition, not executing a merging process; changing the scale coefficient, and continuing to execute the multi-scale segmentation; repeating the previous step until completion of a scale increase process, and forming a multi-scale image segmentation result with image edge intensity as a constrained condition. According to the invention, an edge deletion problem in image segmentation is solved, and a scale selection mode is provided in image multi-scale segmentation.

The invention discloses a hardware circuit implementation method for an image matching descriptor generation scale space. The method comprises a step of loading an image to be matched in a specific application, a step of performing Gaussian smoothing on the image to generate a first group of images, a step of generating a next group of images by using a method of the combination of downsampling and Gaussian filtering, and a step of generating a Difference of Gaussian pyramid according to an obtained Gaussian pyramid. By redesigning a scale space generation method of an SIFT image matching algorithm to adapt to the characteristics of fixed point, parallel and pipeline of an FPGA operation, the speed of scale space generation is accelerated, hardware resources are saved, and therefore, the performance of a whole feature matching module is effectively optimized.

The invention relates to a satellite constellation system taking account of both large-scale space detection and small-scale space detection, which comprises a plurality of satellites and a plurality of relay communication terminals, each satellite further comprises an integrated electronic computer, a spatial magnetic field detection load, a spatial energetic particle detection load and a spatial plasma detection load, the integrated electronic computer is connected with the spatial magnetic field detection load, the spatial energetic particle detection load and the spatial plasma detection load, and the satellites are distributed layer by layer, and communicate with one another through the relay communication terminals. The satellite constellation system can fully meet the requirements of the applications of large-scale solar burst-caused near-earth space environment response detection and small-scale solar burst-caused near-earth space environment response detection and the mounting and operational requirements of detection loads for spatial magnetic field, energetic particles, plasma and the like in the near-earth environment, and has the technical characteristics of integrated design, multi-satellite networking, different-orbit distribution, dynamic configuration and the like.

The invention relates to a multi-scale segmentation-based saliency detection method. The method includes the following steps that: 1: smoothing image processing is performed on an input image through using bilateral filtering parameters, super-pixel segmentation of different segmentation scales is performed on the processed input image, global smoothness is calculated according to super-pixels obtained through segmentation, the global smoothness and the bilateral filtering parameters are combined to build an adaptive algorithm function adopting a segmentation effect as an objective, bilateral filtering parameters under different scales are solved, and super-pixel points in the optimal smoothed image are obtained; sep 2, initial foreground seeds are obtained through using a target likelihood graph technique, the boundary of the image is adopted as initial background seeds, background seeds and foreground seeds are selected from the initial background seeds and the initial foreground seeds by using a cross-validation method, and a background-based RBB saliency map and a foreground-based RFB saliency map are generated; and step 3, the scale weights of the super-pixels and the seed weights of the background seeds and the foreground seeds are calculated, and the obtained RBB saliency map and RFB saliency map are combined, so that a final saliency map can be obtained.

The invention relates to a flotation broken froth detection method based on rapid retina feature point matching and multi-scale segmentation. The flotation broken froth detection method comprises the following steps: firstly, collecting two continuous frames of flotation froth images, carrying out NSST decomposition on the two frames of froth images, carrying out froth edge detection and fusion on a multi-scale high-frequency sub-band, and extracting a central point of each segmented froth in a next frame of image; secondly, carrying out feature point description and matching on the two frames of images by adopting an improved FREAK sampling model, and extracting candidate broken froths according to the distribution density of matching points around the previous frame of segmented froths; finally, mapping the central point of each segmented froth in the next frame of image into the previous frame of segmented image, counting the central point number contained in the candidate broken froths; and judging the candidate broken froths containing a plurality of center points or without center points as broken froths. For the flotation broken froth detection method based on rapid retina feature point matching and multi-scale segmentation, the improved FREAK algorithm is high in matching effect and real-time performance, and the broken froth detection method is less influenced by illumination and motion changes, and the broken froths can be effectively extracted.

The invention relates to an image segmentation method and system. A super pixel is extracted from an image to be segmented, a regional center point of the super pixel is obtained and marked, and mark information of the regional center point of the super pixel is used to mark unmarked sample points. When the regional center point of the super pixel is marked, sparse simplification is carried out on an undirected weighted graph constructed by the sampling points, and a graph based semi-monitoring learning algorithm is used for marking. When the other unmarked sampling points are marked, the unmarked sampling points are marked in a classified manner by utilizing K-neighbor and taking the regional center of the super pixel as basis. The method and system are characterized by being high in speed, high in marking accuracy and suitable for large-scale image segmentation.

The invention discloses a slope unit division method based on multi-scale image segmentation. The method comprises the following steps: obtaining a slope map layer based on a digital elevation model;calculating to obtain a unit vector in the X-axis direction and a unit vector in the Y-axis direction, and taking the unit vectors as main body segmentation layers for dividing slope units; extractinga catchment area map layer and taking the catchment area map layer as a limit map layer for slope unit division; and dividing the slope units to obtain a final slope unit division result. According to the method, topographic elements are comprehensively considered, a data set suitable for multi-scale segmentation, an optimal segmentation scale and a segmentation algorithm are established, and anappropriate slope unit is generated; therefore, the method provided by the invention can provide reliable map unit support for geological disaster risk evaluation, and the slope unit division method provided by the invention is high in reliability and relatively high in slope unit division precision.

The invention relates to a multi-scale segmentation constrained remote sensing image semantic segmentation method, which comprises the following steps: carrying out multi-scale region merging on a remote sensing image, carrying out fusion processing on a merging result to obtain a multi-scale segmentation result, and carrying out semantic prediction through a deep learning network to obtain a remote sensing image ground object accurate boundary and semantic information. According to the method, a multi-scale segmentation result is used as a constraint, so that on one hand, a remote sensing image ground object accurate boundary can be obtained; on the other hand, the semantic information of the closed area can be obtained, and the problems that the semantic information of the closed area cannot be directly obtained through a traditional multi-scale segmentation method and the accurate boundary of the ground object cannot be well reserved through a semantic segmentation method based on deep learning are solved.

The invention discloses a micro calcification point automatic detection method based on an ultrasonic breast tumor image, and the method comprises the steps: segmenting an original image of the ultrasonic breast tumor image to obtain a breast tumor interest region; carrying out weak corrosion and strong corrosion distinguishing on the segmented breast tumor interest region; carrying out multi-scale superpixel segmentation on an ultrasonic breast tumor weak-corrosion interest region, and fusing the texture segmentation result of each scale to obtain a first suspected calcification point with a compact edge; performing single-scale superpixel segmentation on the original image of the ultrasonic breast tumor image, combining the original image of the ultrasonic breast tumor image after the single-scale superpixel segmentation with an ultrasonic breast tumor strong-corrosion interest region, and co-screening out a second suspected calcification point closer to the target based on the gray scale comparison characteristic and the gray scale distance characteristic; and obtaining the micro calcification point of the ultrasonic breast tumor with the accurate target and the compact edge. The method ensures that the detected tiny calcification point is accurate and reliable, is ingenious and novel to realize, and has a good application prospect.

The invention discloses a multi-scale segmentation method, which is for post-processing the maximum expected remote sensing image change detection and comprises the following steps of: step a, carrying out spatial position registration on the front image and the rear image of the landslide in the landslide area to obtain the changed image; step b, carrying out multi-scale segmentation on the basisof the rear image of the landslide, and acquiring the multi-scale segmented image object set S, and step c, acquiring the ith object Oi in the S, performing spatial stacking analysis with the initialchange detection result in step a and respectively counting the number of changing pixels and unchanged pixels in the object Oi; step d, refining the pixel attribute in the object Oi by utilizing themaximum expected algorithm; step e, let j=i+1; if j is less than or equal to n, taking the j object Oj in the S, and sequentially executing the step c and the step d, until j) n; Step f: obtaining afinal change detection result. The method has strong universality and obvious effect, and discloses a multi-scale segmentation method, which is for post-processing the maximum expected remote sensingimage change detection.

The invention relates to the field of multi-scale image segmentation, in particular to a multi-scale image segmentation method based on hierarchical region merging. The method comprises the steps of generating a multi-scale segmentation result by using a multi-scale image segmentation method, extracting a plurality of segmentation results from low to high according to a threshold value, and establishing a multi-scale hierarchical region merging tree; carrying out segmentation quality quantification on the segmentation area of each scale, and carrying out global image hierarchical synthesis byutilizing an optimization algorithm; and determining a set of suitable segmentation regions according to a result of the synthesis hierarchy, and generating an optimal segmentation result of the imageaccording to the set. According to the method, the segmentation precision loss caused by manually setting a threshold in a multi-scale image segmentation method is overcome, and the optimal segmentation scale selection of an individual target can be realized to a certain extent.

The invention discloses an SDN flow rule detection method based on a regional growth algorithm, and the method mainly comprises the steps: determining the dimension of a flow rule space through judging whether an MAC address, an IP address, and a TCP/UDP protocol port are enabled in a flow rule; and searching a flow rule in the flow rule space based on the dimension of the flow rule space by adopting a region growth algorithm, performing space segmentation, taking the subspace obtained by segmentation as a new to-be-detected space to perform circular detection, and taking the flow rule obtained by detection as output to complete SDN flow rule detection based on the region growth algorithm. According to the invention, the detection flow can be greatly reduced, the detection time is shortened, and the concealment is stronger. Moreover, the method does not need to know the structure information of the target network, does not need to have the management and operation authority of the SDNcontroller and other network components, and has relatively high applicability.