471results about How to "Short runtime" patented technology

The present invention provides an S-shaped acceleration and deceleration control method for changing speed and position of an object on line. The method comprises the acceleration phase speed planning, the deceleration phase speed planning, the constant speed phase speed planning, the real speed reduction point predication, the maximum speed processing, the surplus distance compensation, the online object speed changing algorithm and the online object position changing algorithm. An acceleration and deceleration discretization speed planning method is employed and the user input parameters are combined to calculate the operation time of a seven-phase speed planning phase. It is determined whether the maximum acceleration and the maximum speed can reach the criterion or not, the integration problem of a sampling period Ts according to the acceleration/deceleration acceleration, the acceleration/deceleration speed and the final position L after discretization is considered, and the real reachable acceleration/deceleration acceleration, the acceleration/deceleration speed and the feed rate are corrected. The S-shaped acceleration and deceleration control method for changing speed and position of the object on line greatly simplifies an original calculation formula and saves lots of operation time of a computer, and the surplus distance employs a one-time compensation method in the speed reduction process.

The invention discloses a voice enhancing method based on a multiresolution auditory cepstrum system and a deep convolutional neural network. The voice enhancing method comprises the following steps:firstly, establishing new characteristic parameters, namely multiresolution auditory cepstrum coefficient (MR-GFCC), capable of distinguishing voice from noise; secondly, establishing a self-adaptivemasking threshold on based on ideal soft masking (IRM) and ideal binary masking (IBM) according to noise variations; further training an established seven-layer neural network by using new extracted characteristic parameters and first/second derivatives thereof and the self-adaptive masking threshold as input and output of the deep convolutional neural network (DCNN); and finally enhancing noise-containing voice by using the self-adaptive masking threshold estimated by the DCNN. By adopting the method, the working mechanism of human ears is sufficiently utilized, voice characteristic parameters simulating a human ear auditory physiological model are disposed, and not only is a relatively great deal of voice information maintained, but also the extraction process is simple and feasible.

The purpose of the invention aims to provide an object and indoor small scene restoring and modeling method based on RGB-D camera data. In summary, the method comprises: in the RGB-D depth data (point cloud) registering and modeling, integrating the restraining conditions for point-to-face and point-to-projection together so that they are applied to the accurate registering of the sequence depth data (point cloud) obtained by the RGB-D camera; and finally obtaining a point cloud model (.ply format) for an object or a small scene wherein the model is able to be used for object measurement and further CAD modeling. The registering method of the invention considers the speed advantage of the point-to-projection algorithm and integrates the precise advantage of the point-to-tangent plane algorithm. This overcomes the problems with the slow speed and low scene precision in a traditional point cloud registering method, and is capable of finding out the corresponding point of a point on the source point cloud to a target point cloud quickly and accurately so as to realize the cloud splicing of multiple view points.

The invention belongs to the technical field of image processing, provides a multi-view-point computing and imaging method, and aims to acquire accurate and high-precision depth information. According to the technical scheme, the multi-view-point computing and imaging method based on a speckle-structure optical depth camera comprises the following steps of: acquiring a depth map and a color map from human-machine interaction equipment Kinect of Microsoft; detecting the edge of the color map, and expanding gap information until the expanded gap covers the edge, which corresponds to the color map, near the conventional depth map; then performing double-edge filtration estimation on points on the edge of the gap according to depth information and color information; limiting the estimation process by virtue of edge information of the color map; and after the color map and the restored depth map are acquired, converting the depth map into a disparity map, and synthesizing any number of virtual viewpoint maps by a disparity map synthesis method combining a depth image and a color image. The multi-view-point computing and imaging method is mainly applied to image processing.

The invention discloses an advanced persistent threat (APT) attack detection method based on deep belief network-support vector data description. A deep belief network (DBN) is used for feature dimension-reduction and excellent feature vector extraction; and support vector data description (SVDD) is used for the data classification and detection. At a DBN training state, the feature dimension-reduction is performed by using the DBN model after obtaining a standard data set; a low-level restricted Boltzmann machine (RBM) receives simple representation transmitted from the low-level RBM by usingthe high-level RBM so as to learn more abstract and complex representation after performing the initial dimension-reduction, and back propagation of a back propagation (BP) neural network is used forrepeatedly adjusting a weight value until the data with excellent feature is extracted. The data processed by the DBN is divided into a training set and a testing set, and the data set is provided for the SVDD to perform training and identification detection, thereby obtaining the detection result. The attack detection method disclosed by the invention is suitable for the unsupervised attack datadetection with large data size and high-dimension feature, is fit for the APT attack detection and can obtain an excellent detection result.

The invention provides a compressed sensing-based antifriction bearing fault diagnosis method based on a working condition disturbance condition. The antifriction bearing fault diagnosis method comprises signal compression, front fault diagnosis and a far-end signal reconstruction algorithm. The method utilizes a vibration signal of a bearing to perform fault diagnosis. Based on the compressed sensing theory, a measuring matrix is constructed and compression of vibration signals is realized so that the transmission bandwidth consumption of the vibration signals of the bearing is effectively reduced. The on-board fault diagnosis part utilizes the compression reference matrix, the matching pursuit algorithm to realize fault diagnosis under the working condition disturbance condition through the reconstruction matching method. Based on the on-board fault diagnosis, the far-end signal reconstruction can be realized through the reconstruction matching method so that fault diagnosis enhancement and performance assessment of the far end can be realized. The method system is complete and is suitable for the working condition disturbance condition, and is high in the accuracy of fault diagnosis and is high in engineering practicality.

The invention discloses a fast target positioning method and device. The method comprises the following steps: according to an input original image, obtaining a binary image; according to the binary image, carrying out target boundary tracking to obtain a target contour, and obtaining a bounding rectangle of the target contour; and according to target characteristics, carrying out target screeningand positioning. The binary image is obtained by carrying out multi-threshold segmentation, denoising and hole filling of the original image, wherein the denoising and hole filling processing is finished through expansion, corrosion, opening and closing operation of mathematical morphology; the boundary tracking is realized through 4-neighborhood or 8-neighborhood chain code tracking; the targetscreening is realized based on a plurality of constraint conditions, wherein the plurality of constraint conditions comprise an aspect ratio constraint, a maximum and minimum area constraint, a regiongray average value constraint and a heel landing constraint; and the positioning comprises acquisition of azimuth and distance. The fast target positioning method and device are applied to the infrared image field, is high in operation efficiency and accurate in analysis.

The invention discloses a wireless network node dynamic deployment method based on network flow, and the wireless network node dynamic deployment method based on the network flow solves the problem that mobile energy consumption is too large due to blindness of moving of sensor nodes during the changing process of coverage area of wireless sensor network, with mobility, of the sensor nodes. The wireless sensor network node dynamic deployment method based on the network flow comprises the following steps that when an area, to be observed, of the wireless sensor network changes, a heuristic type area coverage optimization method based on a genetic algorithm is utilized to calculate the position where the sensor nodes need to be deployed, then a network flow algorithm is used for planning the moving route of the sensor nodes reasonably according to the all-sensor-node shortest total route principle. The wireless network node dynamic deployment method based on the network flow can calculate the best objective positions to be covered of the wireless sensor network nodes and the shortest moving route of the sensor nodes in the coverage area of the sensor nodes, and reduces moving energy consumption of the sensor nodes in the network coverage deployment.

A therapeutic apparatus for medical applications is improved in such a manner that the capacity of an energy store can be utilized as best as possible when there is a breakdown of the power supply. A charge level detector (16) is provided for the energy store which reduces the power output of the energy consumer (6) when there is a falling charge level of the energy store (14) with a drop of the supply voltage (15).

The invention discloses a fast tracking recognition method for overlapped fruits by a picking robot. The fast tracking recognition method comprises the following steps: continuously collecting the latest ten frames of overlapped apple images through a camera; segmenting the collected first frame of image, and removing a background; determining the position of the circle center of overlapped apples by calculating the maximal value of the minimal distance from points in a circle to the edge of an outline; calculating the distance from the circle center to the edge of the outline to determine a radius; intercepting a subsequently matched template according to the circle center and the radius; determining the circle centers of overlapped apples in the continuously collected latest tens frames of images, and carrying out fitting and pre-judging on the motion path of the robot according to the circle center of each frame of image; determining the positions of overlapped apples in a next frame of image by synthesizing the radius and the pre-judging path, and intercepting the area of the overlapped apples; finally carrying out matching recognition by adopting a rapid normalized cross-correlation matching algorithm. According to the method, tracking recognition of near-spherical overlapped fruits such as the overlapped apples can be achieved; the running time is short; the picking efficiency of the picking robot can be effectively improved.

The invention discloses a power transformer fault diagnosis method based on an improved firefly algorithm (PFA) optimized probabilistic neural network (PNN). The power transformer fault diagnosis method comprises the following steps: firstly, collecting fault characteristic gas by using a gas chromatographic analysis method and carrying out pretreatment by using a fused DGA algorithm; initializinga PNN neural network, a firefly algorithm and a two-dimensional particle swarm; taking the PNN smoothing factor as a firefly individual, and calculating the position and brightness of the firefly; feeding the solving result of each firefly algorithm back to the particle swarm algorithm, carrying out fitness evaluation on each particle, and updating the positions and speeds of the particles; carrying out loop iteration, substituting the obtained optimal smoothing factor into the PNN to carry out fault prediction, and training a PNN model after PFA optimization; inputting a test sample, and outputting a fault type result, thereby achieving the fault diagnosis of the power transformer. The method is high in search speed, high in diagnosis precision, small in error, and obvious in classification effect.

The invention discloses a system and a method for verifying register transfer level (RTL) hardware of a video algorithm. The system comprises a test video sequence library, a test vector generator, a golden C language model, an RTL hardware model to be verified and a file comparator, wherein the test video sequence library is used for storing a test sequence required by verifying the design of the RTL hardware of the video algorithm; the test vector generator is used for selecting the test sequence from the test video sequence library according to the functional coverage of the algorithm, generating a test vector and outputting the test vector to the golden C language model and the RTL hardware model to be verified; the golden C language model and the RTL hardware model to be verified are used for respectively generating output after receiving the test vector and outputting the respective output to the file comparator; and the file comparator is used for comparing whether the output of the golden C language model is consistent with the output of the RTL hardware model to be verified or not, indicating that the RTL hardware passes verification if the outputs are consistent, and indicating that the RTL hardware does not pass verification if the outputs are inconsistent. By the system and the method, the efficiency and correctness of verifying the design of the RTL hardware of the video algorithm are improved.

The invention discloses a data mining method for multi-index evaluation information. The method includes: reading existing interactive data including explicit rating data which include data rated by a user on already rated other objects or services similar to to-be-rated objects or services and data rated by other users on the to to-be-rated objects or services; and according to the existing interactive data, adopting a trained association model to calculate a rated value of the user on a certain object or service in a certain index. The method can support predication of multi-index rating, has high predication accuracy and classification accuracy, and is high in convergence rate, short in operation time and suitable for on-line real-time recommendation.

An iris external boundary positioning method based on shades of gray and a classifier belongs to the technical field of image processing and relates to iris-identification technology. The method includes the following steps of: firstly determining an approximate region of the iris external boundary by pupil boundary information; secondly conducting histogram extension on the region and calculating lash gray threshold; subsequently using the shape and distribution characteristics of the iris external boundary to design two boundary point classifiers and obtain a point set of the iris external boundary; then according to the obtained point set of the iris external boundary, adopting a least square method to fit iris external boundaries ready to be selected; and finally utilizing pupil position information to compare the iris external boundaries ready to be selected and obtain the proximate iris external boundary curve. Compared with the existing iris external boundary positioning method, the iris external boundary positioning method has not only fast speed and accurate positioning, but also high robustness and the capability of still properly positioning an iris image with poor image quality, thus being conducive to improving the identification efficiency and accuracy of iris.

The invention discloses a super-resolution image reconstruction method based on structure self-similarity and sparse representation. The method includes the main steps of firstly, filtering a set of training sample image to extract features; then, extracting small patches to construct a dictionary including a high-resolution image block and a low-resolution image block in pair, conducting interpolation amplifying on an inputted low-resolution image, conducting filtering to extract the features, solving a reconstructed weight matrix W, conducting iteration to renew a sparse coefficient {alpha i} and a high-resolution image X to be reconstructed; finally, recovering a satisfying high-resolution image till the iteration is convergent. According to the method, the structure self-similarity of the image is used for solving the problem that an existing method is not high in quality. The operation time is short, the efficiency of image reconstruction is high, the quality of the reconstructed image is high, and various natural images which include non-texture images such as animal and plant images and human images and strong-texture images such as architecture images can be reconstructed.

The invention discloses a detection algorithm of a cement concrete pavement slab staggering quantity. The detection algorithm specifically comprises the following steps of: inputting a three-dimensional image data matrix, and filtering the data; orderly taking each line of the three-dimensional image data matrix O'' after filtering, and solving the slab staggering quantity corresponding to each line in a line-by-line manner, namely solving the slab staggering quantity of the line after finding out all inflection points of the line; taking the inflection points, of which the difference of the horizontal ordinates is M, from an inflection point matrix G, and solving the difference between each two inflection points so as to obtain a height difference matrix after all the lines are processed, wherein the maximum of the height difference matrix is the slab staggering quantity corresponding to the line; solving the slab staggering quantities h1, h2,... and hm in a line-by-line manner, averaging the m slab staggering quantities, and thereby obtaining the pavement faulting quantity of the image acquisition area. The detection algorithm of the cement concrete pavement slab staggering quantity is simple in computation and short in run time, and does not need manual work. By using the surface measurement, the pavement faulting quantity can be detected by only inputting the acquired three-dimensional image data of the cement concrete pavement; and therefore, the detection algorithm is high in efficiency and accurate in detection.

The invention provides a method for automatically identifying and grading the central segregation defect of the low-magnification structure of a continuous casting billet. The method comprises the following steps: S1, preprocessing the gray image of a low-magnification structure of a continuous casting billet, including distortion correction, cutting white background and filtering; S2: obtaining acentral area image by segmenting the gray image for the central area where the area of the horizontal direction (2/5-3/5) and the area of the vertical direction (2/5-3/5) of the gray image; S3, detecting and labeling the connected region in the binary image corresponding to the central region image to separate the suspected central segregation region from the crack region in the connected region;S4, carrying out feature extraction on that suspected central segregation region, identifying the central segregation region and removing the interference region; S5: training the BP neural network classifier model to identify the center segregation region and grade the same. The technical proposal of the invention solves the problem that the traditional digital image processing method is not suitable for the defect detection of the continuous casting billet whose surface defects are jumbled and liable to cross.

The invention discloses a general steganography method based on deep learning. The method comprises the following steps: S1, carrying out hiding processing on a sender; dividing secret information tobe hidden into n groups of information fragments, wherein each group of information fragments correspond to one category label, a deep learning model is adopted, the category label and random noise are used as drive, a pseudo-natural image of a specified category is generated, and the pseudo-natural image is used as a secret-containing image input channel after hiding processing; and S2, carryingout extraction processing at a receiver: inputting the secret-containing image into a discriminator by the receiver to carry out image authenticity identification and image category judgment, then sending the image category information into a function converter to be processed to obtain a secret information fragment, and decoding the secret information fragment to obtain original secret information. According to the invention, the security and confidentiality of information transmission can be greatly improved.

The invention discloses a grooved texture depth detection algorithm for a cement concrete pavement. The grooved texture depth detection algorithm comprises the following steps: inputting a three-dimensional data matrix of an image, and performing filtering processing on data; performing partitioning processing on the three-dimensional image data matrix after filtering, and dividing the data matrix to N uniform blocks according to the size of the matrix; calculating the corresponding texture depth of the block one by one; and calculating the mean texture depth one block by one block, and calculating an arithmetic mean of the N mean texture depths, namely the texture depth value TD of the image. According to the grooved texture depth detection algorithm, the real-time, high-efficient and quantitative detection of skid-resisting capacity of the cement concrete pavement can be realized. The method provided by the invention has the advantages of simplicity in calculation and short operation time, and is suitable for being used in a real-time system. According to the method provided by the invention, human intervention is not required, surface measurement is adopted, the efficiency of the method is high and the detection is precise; the influence of an inclined plane is eliminated, and the range of applications is wide, and potent information support can be provided for maintenance and management of the cement concrete pavement, and the maintenance and management level of a highway is improved.