183results about How to "Reduce redundant calculations" patented technology

The invention relates to a hybrid image processing method of extracting the centric line of the structured-light stripe based on the ROI area, belonging to the field of machine vision technique. By combining the image threshold processing, communicated components extraction, shape factor constraint and expansion method, the invention automatically divides the structured light area as the ROI area of light stripe extraction, using the extraction method of fringe center based on Hessian matrix to realize the extraction of sub-pixel image space of the centric line of the structured-light stripe of the ROI area. The method has the advantages of high precision, good robustness, high automation degree and other advantages, avoiding the large formwork Gaussian convolution calculation of the pixel the non-light stripe area, thereby reducing the redundant calculation of the extraction of the structured light stripe and realizing the high speed and precision extraction of the centric line of the light stripe, laying a foundation for the real-time application of the structured light 3D vision measurement.

The invention discloses establishment of a virtual case library of cancer pathological images and a multi-scale cancer detection system based on a convolutional neural network. The system is based ona method of a convolution neural network, a cancer mass region is detected on a pathological full-scan section, and the system includes four modules: 1) a pathological section image preprocessing module; 2) a virtual case database construction module; 3) a high-scale cancer mass detection module; and 4) a small-scale cancer mass classification module. The multi-scale cancer detection system provided by the invention can make full use of the multi-scale information of a pathological image, on different scales, according to the characteristics of the image, different strategies are designed to detect a suspected cancer area, and at the same time, under the condition of insufficient training data, the virtual case library method established by the invention can provide more training data setsfor an existing data-driven deep learning method. The multi-scale cancer detection system based on a convolutional neural network has the characteristics of multi-scale detection, driving of a relatively small amount of data and the like, and has the characteristic of reducing computing resources required for one-time recognition and improving time efficiency of an algorithm on the basis of ensuring the overall recall rate and accuracy.

The invention discloses a vehicle target detection method and aims to solve the technical problems that a traditional method is complicated in detection process, large in calculated quantity and inaccurate in detection. According to the method, a VOC 2007 vehicle dataset is made according to an ImageNet dataset; a training model is configured by the adoption of a caffe deep learning framework andan improved Faster R-CNN algorithm, and an Inception network structure is introduced to perform feature extraction on an image; a sliding window with the area of 642 is additionally arranged and usedfor detecting a small target; a target detection problem in a scene is converted into a target dichotomy problem, and a vehicle target is detected and recognized; an RPN loss function is used to perform optimization; and a SoftMax algorithm is utilized to classify vehicle image features, and therefore a final result is obtained. The method has the advantages that the calculated quantity of the image features is reduced, the detection process is simplified, the image feature extraction capability is enhanced, network recognition precision is improved, and the detection result is more comprehensive.

The invention discloses a thermal image defect detection method of a pressure vessel based on line variable step size segmentation. The method comprises the following steps: selecting a maximum pixelvalue point from eddy current pulse thermal image, computing a related coefficient by utilizing a Pearson correlation coefficient method according to the maximum pixel value point, and then obtainingthe conversion step size; computing the transient thermal response of each pixel point by combining the conversion step size, and performing classification processing; and performing defuzzfication onall pixel points by utilizing the membership maximization principle to obtain the category of each pixel point; and finally performing feature extraction by adopting a Canny operator algorithm afterperforming dimension-reduction processing, thereby extracting the defect feature in the eddy current pulse thermal image.

The invention discloses a low-decoding complexity rate matching polarization code transmission method based on a QUP method, and belongs to the technical field of the channel coding / decoding. The method comprises the following steps: step one, determining a punch position and a punch parameter; step two, constructing an information sequence; step three, performing polarization coding, and outputting a coding sequence; step four, outputting the punched coding sequence according to the punch parameter determined in the step one; step five, sending the coding sequence punched in the step four one bit by one bit by a sending end; step six, receiving the coding sequence sent in the step five through channel transmission by a receiving end; step seven, optimizing a decoder structure according to the punch parameter determined in the step one, and outputting the optimized SC decoder; and step eight, performing polarization code decoding by using the SC decoder optimized in the step seven. By using the method disclosed by the invention, the spatial complexity and the time complexity of the polarization code decoding are effectively lowered, and the saved time and time source is increased along the increasing of the punch number.

The invention discloses a rail defect segmentation self-adaptive fast algorithm based on two-dimensional Otsu, which comprises establishing a two-dimensional Otsu segmentation direct division map, dividing an image to be segmented into a target region and a background region, and setting an adaptive threshold search interval. A pixel point of a threshold retrieval area is partitioned, for the graylevel of each region, Compare the mean variance of each region, find out the region where the best threshold point is, then calculate the inter-class variance of each pixel gray value in this region,and get the maximum inter-class variance value, the corresponding gray value is the best segmentation threshold value, according to the best segmentation threshold value, the image is binarized intothe target region and the background region. When the rail defect image is segmented, the threshold search interval is adaptively set to reduce redundancy calculation in the threshold search process,meanwhile, the threshold interval narrows the gap between the background and the target area, reduces the interference caused by the noise point to the threshold determination, and solves the problemthat the defect in the small area is difficult to be segmented.

The invention discloses a collaborative virtual and actual sheltering treatment method in a shared enhanced real scene. The collaborative virtual and actual sheltering treatment method comprises the following steps of: generating an enhanced real scene; estimating the position of a point to be observed; projecting both a virtual object and a real object on an observation plane of the point to be observed, wherein the sheltering relationship between the virtual object and the real object is to be analyzed; taking the point to be observed as an initial point; taking a ray vertical to an imaging plane to scan the observation plane along the horizontal direction; carrying out sheltering encoding according to a successive sequence that the ray touches two sides of the virtual object and two sides of the real object when the ray scans the observation plane; determining whether the virtual object and the real object which are observed from the point to be observed are in either a sheltering relationship or a non-sheltering relationship according to the difference among sheltering codes; and specifically judging that the virtual object shelters the real object or the real object shelters the virtual object so as to reduce wrong judgment. According to the collaborative virtual and actual sheltering treatment method in the shared enhanced real scene, the quick judgment of the spatial sheltering relationship between the virtual object and the real object in a new video sequence region is achieved, and the accuracy in the judgment of the virtual and actual sheltering relationship in the enhanced real scene is improved.

The invention discloses an image style migration method combining a meta-learning mechanism and feature fusion. According to the image style migration method, feature fusion based on convolution calculation and a method for decoding a feature map by using the meta-learning mechanism are combined. Firstly, content features and style features are preliminarily fused through convolution calculation,weighted summation operation is carried out on a preliminarily fused feature map and a preliminarily fused content feature map, and the stylization degree is controlled by adjusting the weight; and then the fused feature map is decoded into a stylized image by using a meta-learning mechanism, and secondary learning is carried out on the style in the decoding process, so that full expression of style features is ensured. According to the invention, the quality of the stylized image is improved, so that the style of the synthesized image is fairer than the original style; controlling the degreeof stylization based on the characteristics of the content image and the style image; a meta-learning mechanism is used to simultaneously carry out style secondary learning and feature map decoding operation so that style migration time is shortened and stylization of any image is rapidly realized.

The invention provides a stereo matching optimization method of a binocular visual system. The method comprises a step of using the binocular visual system to gain a binocular image; a step of correcting a distortion and polar line of the binocular visual system; a step of transferring the binocular image from a colorful image into a grayscale image; a step of regarding one of the grayscale images as a basic standard and matching with another image according to pixel masterplate; a step of optimizing ZNCC factor; a step of regarding the optimized ZNCC as a similarity measure to calculate in a matched mode so as to ensure matching parallax of each pixel point; a step of producing a dense disparity map according to the matching parallax of each pixel point. The method has the advantages of reducing times of calculation greatly and reducing calculating load of computer chip greatly by optimizing a computational formula of the ZNCC factors and calculating method. The method can improve real-time performance of the stereo matching as well as reducing cost of day-to-day and military high-precision image identification.

The invention provides an acceleration method and device for a convolutional neural network, and relates to the technical field of image processing. The method comprises the steps that a semi-stop module is arranged in the convolutional neural network; in the prediction process of the convolutional neural network, a current prediction result of the prediction process is calculated when the semi-stop module is executed; whether the current prediction result meets preset prediction conditions or not is judged; when the current prediction result meets the preset prediction conditions, the prediction process stops, the current prediction result is taken as a final prediction result of the convolutional neural network, and otherwise, the prediction process continues to be executed. According to the acceleration method and device for the convolutional neural network, the technical problems that in the prior art, the operation speed of the CNN is low, and tasks which have a high requirement on the real-time performance are difficult to execute can be solved.

The invention provides a real-time medical ultrasonic three-dimensional imaging method, which is applicable to real-time acquisition of three-dimensional medical ultrasonic volume data which may be stored. Aiming at the characteristics of a current central processing unit (CPU) and the characteristics of a three-dimensional ultrasonic image, storage and access modes of three-dimensional data are optimized, and the conventional ray casting method is improved in combination with the knowledge of image morphology, so that the reconstruction and visualization of the three-dimensional data are quickly realized.

The invention discloses a method for detecting thermal imaging defects of a high pressure vessel on the basis of variable step size segmentation of lines. The method comprises the following steps: selecting the maximum pixel value point from an eddy current pulse thermal image, calculating a correlation coefficient according to the maximum pixel value point by utilizing a Pearson correlation coefficient method, and further obtaining the transformed step size; calculating transient thermal response of each pixel point in combination with the transformed step size, performing classification processing, and performing defuzzification on all the pixel points by utilizing a membership degree maximization criterion so as to obtain the category to which each pixel point belongs; finally, performing feature extraction after dimensionality reduction by adopting a Canny operator algorithm, so as to extract the defect features of the eddy current pulse thermal image.

The invention provides a decoding method of a convolutional Turbo code for reducing decoding time delay and saving a memory. The decoding method comprises the following steps of: simultaneously carrying out forward recursion and backward recursion in a component decoding process; dividing the forward recursion and the backward recursion into two stages with equivalent computation quantity; and sequentially calculating and obtaining posterior likelihood ratio information at the beginning of the second stage. The time delay from the beginning of recursion operation to the end of the posterior likelihood ratio information operation is shortened once compared with the traditional decoding process. Furthermore, the traditional posterior likelihood ratio operation is serial, while the posteriorlikelihood ratio operation of the invention is carried out bidirectionally and simultaneously in parallel, the required calculation time and the recursive calculation time are overlapped, and it is unnecessary to distribute additional calculation time; in addition, a bidirectional parallel structure can ensure that the memory used for storing state metric is reduced by half. Furthermore, through the calculation of splitting branch metric, redundancy calculation is reduced, and the space for storing the branch metric is reduced by half.

The invention discloses a method for verifying the integrity of data in hybrid cloud storage. The method is characterized in that based on a PDP algorithm, a secret key is distributed for the file anda label is generated by means of a chameleon signature algorithm and a private cloud and is stored in a public cloud; when the selected file is verified, the private cloud generates verification information and sends the verification information to the public cloud, the public cloud obtains the secret key and the label to form evidence, the private cloud serves as a platform, the evidence collides with the to-be-verified file block, and whether the to-be-verified file block is complete or not is judged. According to the data integrity checking method and device, by means of chameleon hash andpublic cloud and private cloud architecture, splitting of the label generation stage in the data integrity checking scheme is achieved, the online waiting time of a user is shortened, the user experience is improved, redundant calculation in the dynamic updating stage is reduced, and the calculation overhead is reduced.

The invention discloses a method and a device for mining an association rule based on attribute reduction. The method comprises the following steps of S1, according to a studying object, collecting corresponding data from a database; S2, performing integration reduction on the data; S3, performing attribute reduction on the data after integration reduction by a rough set theory; S4, mining the association rule on the data after attribute reduction by an improved HEclat algorithm; S5, interpreting and showing the mined association rule. The method has the advantages that the attribute reduction is performed by the rough set theory during data processing, so as to avoid using a large amount of unnecessary attributes for interaction operation and increasing the computing of data quantity; in the data mining process, the data mining of the association rule is performed by the improved HEclat algorithm, and the database is only scanned at one time, so that the redundancy computing is reduced, the data mining time is effectively shortened, the consumption of system resources is reduced, and the data mining efficiency is improved.

The invention relates to a non-contact video displacement measuring method. The method comprises the following steps that 1, motion videos of a digital equipment photographing structure are adopted and guided into a processor; 2, the processor obtains a first frame image in the motion videos, a target point is adopted as a center selection target area; 3, the next frame image is obtained, a target point in the previous image is adopted as a center selection matching area, pixel points in the matching area obtain matching blocks equal to the size of the target area by adopting the pixel point position as the center; 4, the target area and the matching blocks are matched to determine a matching block matched with the target area, the center point of the matching blocks is the position of the target point in the frame image, the position is adopted as a novel target point, and the matching blocks are adopted as the novel target area; 5, the steps 3 and 4 are repeatedly executed to obtain the position where the target point in each frame image is located, and a displacement curve of the target point is obtained. Compared with the prior art, measuring equipment is cheap, and the measuring result is reliable.

The invention discloses a network communication quintuple (a source IP, a target IP, a source port, a target port and a protocol number) fast matching algorithm based on an improved automatic state machine. According to the algorithm, a quintuple unit dividing module, a mixture automatic state machine module, an asterisk wildcard mapping module and a matching rule calculating module are adopted, wherein the quintuple unit dividing module is used for establishing a mismatching collision domain, the mixture automatic state machine module is used for achieving uniform matching of point data and section data, the asterisk wildcard mapping module is used for solving the asterisk wildcard matching problem in matching, and the matching rule calculating module is used for reducing redundancy operation caused by asterisk wildcard mapping. The algorithm is characterized in that static parameters in the quintuple are extracted as mush as possible to establish an ordinary automatic state machine, and a larger collision domain is established; for section type parameters, a chain table is added behind the automatic state machine, a mixture automatic state machine structure is formed, and supporting for the section type parameter matching is achieved; the asterisk wildcard matching problem is solved through asterisk wildcard mapping, a replacement rule is precisely matched, and the amount of redundancy calculation is reduced. The algorithm can be widely applied to an intrusion detection system, a network blacklist and whitelist library, a network data analyzing product and other products.

The invention relates to a digital circuit, and discloses a memory optimization type static time sequence analysis method and system. According to the method, on the premise that the subgraphs are notexpanded, in the traversing and scanning process, false paths are identified and discarded in real time, the influence of the false paths on downstream nodes is reduced, redundant calculation is reduced, meanwhile, multiple time sequence constraints on each node are dynamically and differentially processed in the traversing process, dichotomy coverage is correspondingly carried out, and the number of labels on each node is reduced. According to the application, occupation of system memory space can be greatly reduced, the system operation efficiency is improved, and the system performance isimproved.

The invention discloses a lung parenchyma CT image segmentation method based on a weighted full convolutional neural network, and belongs to the field of medical image processing. The method comprisesthe following steps: selecting a public lung data set for preprocessing, and extracting a lung parenchyma boundary in a labeled image as a semantic category; designing an improved network structure based on a standard full convolutional neural network framework, and establishing an overall structure framework of the pulmonary parenchyma segmentation convolutional neural network by using a principle that a standard path structure for encoding and decoding simultaneously comprises jump connection, expansion convolution and batch normalization; adopting a weighted loss function layer; dividing the data set; carrying out offline model training out to acquire model weight parameters; inputting a test image and outputting a segmentation result by an output layer through layer-by-layer feedforward of a network. According to an existing lung parenchyma segmentation method, a segmentation missing phenomenon is prone to occurring in a focus area in lung parenchyma, and correct segmentation of the focus area in lung parenchyma segmentation can be effectively improved through enhancement processing on important pixels.

The invention discloses a similarity connection query method and device, and relates to the field of data processing. The method comprises: when similarity connectivity query is carried out; The method comprises the steps of firstly obtaining an original vector set for similarity connection query; the number of vector pairs of the similarity connection query results and an initial data set of thesimilarity connection query results are determined; performing grouping processing on the original vector set; obtaining a plurality of sub-vector grouping sets; constructing a similarity distributionhistogram of the original vector set; according to a similarity distribution histogram and the number of result vectors, calculating a similarity distribution histogram; calculating vector distance thresholds, finally, grouping the sets according to a plurality of sub-vectors; and the initial result vector pair set is updated according to the vector distance threshold value and the result vectorquantity to obtain a result vector pair set for representing the similarity connection query result, so that the vector distance threshold value does not need to be set manually in advance, a large amount of redundant calculation can be reduced, and the similarity connection query efficiency is improved.

The invention discloses a power load clustering analysis method, device and equipment. The method comprises: calculating the shortest bending path of the daily load curve data of the power consumer through a dynamic time bending path algorithm; the minimum average distance is calculated, the data representation of the power consumer is determined, effective integration of historical load data of the power consumer is completed, redundant calculation of repeated information is reduced, calculation expenditure is effectively saved, and the problem that the performance of a traditional clusteringalgorithm is reduced due to time fluctuation and uncertainty of the load is solved to a certain extent. According to a preset fuzzy margin, compression dimension reduction processing is carried out on the data representation set, so that the calculation dimension is greatly reduced, and the calculation efficiency is improved. Therefore, the method provided by the invention solves the technical problems of high calculation dimension by using the original daily load data, information redundancy, low calculation efficiency and large calculation result error caused by time uncertainty and volatility in the existing power load clustering analysis method.

The invention provides a Bessel integral adaptive segmentation method and system in rapid calculation of an integrated circuit. The method comprises the following steps: determining the order of a Bessel function used in a Holographic Green function; calculating a zero point of the Bessel function by adopting an iterative method according to the order of the Bessel function; determining the range or the number of the zero points; adaptively segmenting the mth sub-interval, calculating the accumulation of the Bessel integral after the segmentation of the mth sub-interval, and accumulating the accumulation result to the Bessel integral of the whole integral sub-interval; and if the conditions are met, calculating a union vector Green function of a field generated by the source point of the current integrated circuit at the field point. The system comprises an order determination module, a zero point determination module, an integral subinterval setting module, a subinterval segmentation module, a subinterval judgment module and a field calculation module. According to the method, unnecessary redundancy calculation in the Bessel integral adaptive segmentation method is reduced, and the Bessel integral of the whole interval is ensured to be accurate.

The invention discloses a collision detection optimization method based on curvature point clustering and a decision tree. The method includes the following steps that: step 1, a hierarchical boundingbox is selected based on curvature point clustering and geometric features; and step 2, the establishment of a hybrid hierarchical tree is optimized based on a decision tree model. According to the method, the curvatures of the contour points of different types of collision objects are calculated, the geometric features of the contour points of the collision objects are analyzed; a K-means clustering algorithm is used to select the appropriate bounding box, so that the degree of matching between the bounding box and the collision objects can be improved; and a Boosting algorithm is used to improve the accuracy of the decision tree model, redundant calculation in an original collision detection algorithm is omitted, the establishment process of the hybrid hierarchical tree is optimized, and therefore, the efficiency of collision detection can be improved.

The invention discloses an improved integral cycle fuzziness search method, relates to the satellite navigation positioning technology field and aims at a problem of the relatively large time consumption because of relatively poor floating point fuzziness solution precision existing in an SEVB algorithm in the prior art. According to the improved algorithm, through limiting the initial search space size and optimizing the calculation process, the number of ambiguity search candidate points and unnecessary redundancy calculation can be effectively reduced, thereby the search time is reduced, and ambiguity resolution efficiency is improved. Compared with the SEVB algorithm, the improved algorithm has more obvious efficiency advantages in the low precision high-dimensional ambiguity search aspect, and ambiguity resolution efficiency can be integrally guaranteed to be relatively more stable and reliable.

The invention discloses a thermal imaging detection method based on row-column variable-step segmentation and a region growing method. According to the method, a maximum pixel value point is selectedfrom an eddy current pulse thermal image, a correlation coefficient is calculated by use of a Pearson correlation coefficient method according to the maximum pixel value point, and then a variable step is obtained; then, a transient thermal response of each pixel point is calculated in combination with the variable step, and classification processing is performed; next, all the pixel points are subjected to defuzzification according to the membership maximization criterion, and the category of each pixel point is obtained; and last, the region growing algorithm is adopted to perform feature extraction after dimension reduction processing, and therefore defect features of the eddy current pulse thermal image are extracted.

The invention discloses a biological sequence analysis platform which comprises a computer, an input module, an analysis module and an output module, wherein the computer comprises a user interface; the input module is used for manually inputting a biological sequence to be analyzed on the user interface, or uploading a readable-format file recording the biological sequence; the analysis module is used for analyzing the biological sequence obtained by the input module to obtain an analysis result; and the output module is used for displaying the analysis result obtained by the analysis module on the user interface, and / or exporting and storing the analysis result in a pre-defined external file outside the biological sequence analysis platform. The biological sequence analysis platform disclosed by the invention has a simple interface, realizes one-stop analysis, and remarkably saves labor and material resources; meanwhile, by adding a batch processing module, the result visualization and storage are enhanced to facilitate the subsequent analysis and view; and in the analysis process, the powerful search algorithm is improved, and the corresponding analysis time is shortened.

The invention discloses a three-dimensional scene reconstruction method based on a GPS information video. The method comprises the following steps that 1, an FOV model set based on GPS information is constructed and redundancy frame data are removed; 2, a reconstruction frame matrix is constructed; 3, a key frame is extracted; 4, a target scene is reconstructed in a three-dimensional mode. According to the three-dimensional scene reconstruction method based on the GPS information video, the video key frame can be rapidly and accurately extracted, and therefore the precision and efficiency of the three-dimensional scene reconstruction method are improved.

The invention discloses a kernel sparse representation-based fast remote sensing target detection and recognition method. The method includes the following steps that: S1, four RGB characteristic channels are created; S2, the four-phase Fourier transformation of the four characteristic channels of a given image is calculated, a phase spectrum is extracted, the images of the four characteristic channels are reestablished through inverse Fourier transformation, and a saliency map can be generated; S3, binaryzation division is performed on the saliency map obtained in the step S2, and candidate regions of interest are extracted; S4, a search box is scanned through an effective sub-window search algorithm, so that image blocks to be detected are obtained, so that a remote sensing target image block training set is obtained; S5, SIFT features are extracted from the remote sensing target image block training set, and a sparse dictionary is generated; S6, a spatial pyramid is adopted to map the SIFT features; S7, kernel sparse representation is obtained; S8, the kernel sparse representation is solved; S9, the space pyramid vector representation of a target is performed; and S10, a linear support vector machine classification algorithm is used in combination to complete a recognition task.

The invention discloses a block content classification method based on a convolution neural network. The method comprises steps of constructing a data set and using the content type of the data set asa tag of a training sample; constructing a convolution neural network; converting the training sample into a gray scale graph; representing each pixel of the gray scale graph by using the eight-bit binary number; extracting the last bit of each pixel to serve as input of the convolution neural network; obtaining a last bit-convolution neural network model through the training; when predicting theinput N*N code blocks, predicting the content type of the current code blocks by use of the last bit-convolution neural network model, and if the output is a camera shooting block, acquiring a classification result; and if the output is a computer generation block, carrying out prediction by use of the last bit-convolution neural network model, and acquiring a classification result of corresponding computer generation text blocks or computer generation non-text blocks. According to the invention, accuracy of prediction of the content type and the calculation efficiency can be improved; redundant calculation can be reduced; and compression quality is improved.

The invention relates to a face recognition method based on an elastic context relation loss function. According to the method, how to overcome the influence of massive unbalanced data on face recognition training is studied; according to the method, the combined data blocks can be effectively mined, useless redundant calculation is reduced, the whole training process is accelerated, the designedtarget function based on the elastic context does not need to introduce extra training parameters for each class, the influence of long tail classes in large-scale training is reduced, and meanwhile,the face recognition accuracy can be improved through soft spacing.