61 results about "Bipartite graph matching" patented technology

A word alignment apparatus includes a word extracting portion that extracts each word from an example sentence and from a translation sentence thereof, an alignment calculator that calculates at least one of a similarity degree and an association degree between a word in a first language and that in a second language to perform an alignment between words respectively included in the example sentence in the first language and those included in the translation sentence thereof in the second language on the basis of a calculated value, and an optimization portion that optimizes the alignment by performing a bipartite graph matching.

The invention discloses a method for matching envelope images. The method comprises the following steps of: respectively representing two envelope images to be matched by using graphs so that the calculation of the degree of similarity between the images is changed into graph matching, and calculating a distance between two graphs by using a bipartite graph matching algorithm based on the minimum weight value. The method comprises the specific steps of segmenting the envelop images, and constructing the graphical representation of the images based on a segmentation result, wherein each vertex of the graph represents an area in the envelope image, and each edge of the graph is used for representing an adjacency relation between two areas. Because the graphical representation of a plurality of images which correspond to the same envelope can be different under the influence of factors such as noise in the acquisition process of the images, a non-accurate graph matching algorithm is adopted in the method. Lots of experimental results show that: the method has relatively higher robustness in illumination, inclining, rotation and the like; and letter information query based on the envelope images can be efficiently realized.

The invention provides a sensor data processing and cooperative prediction method based on edge calculation in the technical field of data processing and edge calculation. The method comprises the following steps: S1, establishing a scene model of a multi-sensor group, a multi-MEC server and an aggregation node; S2, the sensor group is configured with a data node with storage and communication functions to store record values of a plurality of past moments, and abnormal value detection and correction are carried out on data; S3, unloading the task data to the MEC server by the data nodes through bipartite graph matching; S4, performing clustering and fitting by the MEC server by using the received data, and sending the parameters to the aggregation node; and S5, receiving input data of theprediction task and performing comparative analysis on the data and the received parameters by the aggregation node; according to the method, the data node is established at the multiple sensors to unload the data cleaning task, and the data are predicted at the aggregation node, so that the accuracy and robustness are improved, and the time delay and energy consumption are reduced.

The embodiment of the invention relates to the technical field of electric vehicle scheduling, in particular to a vehicle scheduling method and device. According to the method, a weighted penalty matrix can be generated based on multiple pieces of obtained order information and multiple pieces of vehicle information; Weighted penalty matrix, an optimal order assignment scheme is found by adoptinga bipartite graph matching algorithm; The embodiment of the invention further provides a vehicle scheduling method and device, for the situation that a plurality of waybills need to be delivered in the order information, the method can carry out delivery route planning, and then accurate and efficient scheduling of vehicles is achieved.

The invention provides an information recommendation method and system based on multi-path optimization matching. Historical payment information, power utilization information, recommendation serviceinformation and service recommendation information of each client are acquired; based on the payment behavior data and the power consumption behavior data, the data is clustered by using a fuzzy C-means clustering algorithm, clustering features suitable for analyzing customer behaviors from the data are selected, customer behavior theme tags are constructed, and different types of power customer behavior mode portraits are achieved; the electricity consumption behavior data and the payment behavior data of the client are clustered, behavior characteristics are analzyed, behavior label characteristics are set, business characteristic attributes are described, a business label set is formulated, and business portrait construction is completed; according to the method, a bipartite graph of aclient label set and a service label set is constructed, a Hopcroft-Karp algorithm with the maximum bipartite graph matching is applied to matching of clients and services, the matched clients are found, and service information recommendation is completed.

The invention provides a cross-platform association detection method for firmware vulnerability, which improves the accuracy and efficiency of the existing cross-platform firmware vulnerability association detection method, combines kNN with SVM, and achieves a better compromise between accuracy and efficiency. Based on the function attribute control flow graph, the bipartite graph matching algorithm is used to transform the function matching problem into the graph structure similarity computation problem, and the penalty term is added to the original algorithm, which improves the accuracy ofaccurate matching. The weighted Euclidean distance and the weighted Markov distance are used to improve the detection efficiency.

The invention provides a PCB image color migration device and method based on clustering analysis, and the device comprises an image collection module which obtains a source image and a target image;a RGB color image acquisition module which is used for respectively acquiring RGB color images converted from the source image and the target image; a color space conversion module which converts theRGB color image into an LAB color image; a color clustering module which is used for setting and clustering the LAB color images by utilizing the number of color clusters to generate a color cluster category list; a color cluster matching module which performs PCB bipartite graph matching on the list of the source image and the target image to obtain a matching result; a color mapping module whichcalculates a transformation matrix according to a matching result, and maps the LAB color image to a new color according to the matrix; and a color space inverse transformation module which convertsthe mapped color into an image after RGB color migration. According to the device, colors in an image are divided into different categories by clustering, color migration is carried out on the colorsof the different categories respectively, and a good image registration effect is achieved.

Embodiments of the invention provide an image style brush realization method and device. The method comprises the following steps of: obtaining matched corresponding points of an input image and a reference image according to density correspondence between the input image and the reference image; determining a first superpixel set of the input image and a second superpixel set of the reference image according to the matched corresponding points, and further determining a superpixel bipartite graph; carrying out bipartite graph matching on the superpixel bipartite graph so as to determine mutually matched superpixel pairs in the first superpixel set of the input image and the second superpixel set of the reference image; and carrying out color conversion on superpixels of the input image according to a first color space and the superpixel pairs so as to obtain a result image. According to the method and device, style conversion is carried out on the input image according to multiple style elements of the reference image so as to obtain the result image, so the multiple style elements in the reference image are directly and completely converted onto the input image, the visual effect related to the color, greyscale and contrast ratio of the input image is changed, and then the result image further has the style of the reference image.

The invention discloses a bipartite graph based in-vehicle network distributed storage method. According to the bipartite graph based in-vehicle network distributed storage method, a distributed storage problem is firstly modeled, and bipartite graph matching is performed, so that optimal in-vehicle network distributed storage of in-vehicle request identification information sent by each in-vehicle node can be achieved under different conditions, and in-vehicle network can respond to maximum in-vehicle request identification information; subsequently, repeated network information stored by roadside units is cleared, resource waste generated by the fact that multiple roadside units respond to the same in-vehicle request identification information is avoided, and meanwhile, in-vehicle request identification information which is met is not affected; finally, in-vehicle request identification information which is not met is collected, and secondary allocation is performed on spare storage space obtained by clearing the roadside units until each roadside unit has no spare storage space, or all the in-vehicle request identification information received by the roadside units is responded, or remaining in-vehicle request identification information cannot be met; a storage resource utilization rate and a data response data are increased, and data service qualities of an in-vehicle network are guaranteed.