水槽滤网

尘桶盖

Salient target detection method and system, electronic equipment and storage medium

The invention provides a saliency target detection method and system, electronic equipment and a storage medium, and belongs to the technical field of image detection and data processing, and the method comprises the steps: obtaining an original image; generating a superpixel image from the original image according to an SLIC algorithm, and extracting a multi-view feature space matrix Y; using the multi-view feature space matrix Y, the transformation parameter matrix P, the potential representation learning matrix U and the similarity affinity matrix S to construct a minimization objective function model for saliency detection of the original image, and solving the transformation parameter matrix P and the potential representation learning matrix U; the transformation parameter matrix P and the potential representation learning matrix U are multiplied and then fused with diffusion models of the foreground and the background, and saliency detection of the original image is completed. According ...

Culture condition control method and system for improving optical attenuation performance of microorganisms

The invention provides a culture condition control method and system for improving the optical attenuation performance of microorganisms, and belongs to the technical field of microbial materials.The method comprises the steps that a target wave band influencing the optical attenuation performance of the microorganisms is obtained; determining a particle size optimization target, a culture medium liquid-solid ratio, a culture environment condition, an inorganic salt addition amount and a spore inoculation amount on a culture medium of the microorganisms in a target wave band; acquiring the lipid content and protein content of the microorganisms in the target wave band to determine the types and contents of a nitrogen source and a carbon source of the culture medium and a microbial spore drying method; and culturing the microorganisms according to a particle size optimization target, a liquid-solid ratio of a culture medium, culture environment conditions, an inorganic salt addition amount ...

Distributed unmanned aerial vehicle cluster parameter estimation method based on adaptive data selection

The invention provides a distributed unmanned aerial vehicle cluster parameter estimation method based on adaptive data selection, and the method comprises the steps: building a cross matching mechanism based on fusion data transmitted at a previous moment, and then obtaining a new data selection factor lambda k (t) generation strategy, and determining whether the unmanned aerial vehicle k exchanges data with the neighbor unmanned aerial vehicle based on the new lambda k (t), thereby improving the estimation precision, reducing the calculation cost and the communication cost, and saving the energy consumption of the unmanned aerial vehicle cluster. The distributed ADS-DMLS algorithm is designed based on a cross matching mechanism, the parameter estimation precision of the distributed unmanned aerial vehicle cluster is improved, the calculation cost and the communication cost of the unmanned aerial vehicle nodes are reduced, and a new information fusion technical scheme is provided for the ...

Method and system for determining priority of feature samples based on significant feature variation

The invention provides a method and system for determining the priority of feature samples based on significant feature variation, and relates to the technical field of data processing. According to the key pixel point determination method based on high-frequency information and strong gradient feature fusion, the union set of the high-frequency information pixel position and the strong gradient information pixel position is taken, and the position of the key pixel point is determined; according to the key neuron determination method based on the activation value difference value, the position of a key neuron is determined through the change of an activation value after an original sample and a variation sample are input; on the basis of amplification of sample and model variation, the number of usable samples is increased through variation operation on key pixel points, and sample combination features are formed, so that the accuracy of feature judgment is improved.

Image fusion model security detection method based on conditional generative network

The invention provides an image fusion model security detection method based on a conditional generative network, and relates to the technical field of data security. The method comprises: obtaining an original model to be subjected to security detection, the original model at least comprising an image fusion model, constructing a replacement model of the image fusion model based on a conditional generative network, and obtaining an equivalent model of the original model by using the replacement model; and determining a suspicious back door of the equivalent model by searching a trigger, and determining a malicious back door of the equivalent model based on a divergence function of the suspicious back door as a security detection result of the original model. According to the invention, the problem that the image fusion intelligent identification model based on the traditional image fusion method cannot be tested by using the existing model security test method due to the separation of ...

APT software family identification method and device and electronic equipment

The invention relates to the technical field of APT recognition, in particular to an APT software family recognition method and device and electronic equipment. The method comprises the following steps: extracting operation characteristics of each piece of known APT software; according to the operation characteristics of the known APT software, an Apriori algorithm is adopted to calculate association rules between the known APT software; obtaining an APT family affirmation result according to an association rule between the known APT software; and identifying the APT family to which the unknown APT software belongs according to the APT family identification result. In this way, information such as behavior characteristics, sources and defense methods of the APT software can be quickly determined by affirming APT software families, and the efficiency of processing illegal behaviors such as network APT attacks and penetration is improved.

Single-pixel imaging asymmetric encryption method based on 3D Arnod transformation

The invention provides a single-pixel imaging asymmetric encryption method based on 3D Arnod transformation, and belongs to the technical field of image encryption. The method comprises the following steps: carrying out 3D Arnod transformation on a one-dimensional light intensity signal collected by a single-pixel detector, carrying out encryption processing on parameters of the 3D Arnod transformation, and transmitting the encrypted parameters of the 3D Arnod transformation through a public channel; wherein the public key and the private key which are subjected to encryption processing are generated through an elliptic curve cryptography mechanism.

Cleaning robot carpet identification method

The invention provides a cleaning robot carpet identification method, a cleaning robot comprises a sleeve and an ultrasonic sensor, the ultrasonic sensor is fixed in the sleeve, and the identification method comprises the following steps: setting a standard echo signal of a normal ground; controlling the ultrasonic sensor to vertically transmit an ultrasonic signal to the current ground, and receiving an actual echo signal reflected by the current ground; and judging whether the actual echo signal is different from the standard echo signal or not, and if so, identifying that the current ground is a carpet surface. According to the method, the accuracy and precision of carpet identification can be improved.

Method for reducing conduction loss and switching loss in driving circuit and driving circuit

A method for reducing conduction loss and switching loss in a driving circuit and the driving circuit are used for a driving circuit and a switching circuit, the driving circuit comprises an analog-to-digital converter, and the switching circuit comprises an inductor and is coupled with a load. The method comprises: performing an analog-to-digital conversion on a load voltage of the load at a first rate; and generating at least one current pulse flowing through the inductor at a second rate; wherein each of the at least one current pulse is completed within a second period corresponding to the second rate; all the at least one current pulse is completed in a first period corresponding to the first rate; a first length of the first period is twice longer than a second length of the second period. According to the method, the pulse width modulation signal of the DC-DC converter can be well arranged and controlled, so that the balance between conduction loss and switching loss is achieved, ...

Vulnerability duplicate removal method and device, equipment and storage medium

The embodiment of the invention provides a vulnerability duplicate removal method and device, equipment and a storage medium, and is applied to the technical field of network security. The method comprises the following steps: acquiring data field information of vulnerability information; according to a preset normalization algorithm, performing normalization processing on data corresponding to the data field information; encrypting data corresponding to the normalized data field information by adopting a digest algorithm to obtain an encrypted character string text; generating a vulnerability number according to the character string text; and performing bug duplication removal according to the bug number. In this way, the recognition precision of repeated weak point recognition can be improved, and therefore the efficiency of vulnerability duplicate removal is improved.

Image generation method and system based on three-dimensional point cloud

The invention discloses an image generation method and system based on three-dimensional point cloud, and relates to the technical field of three-dimensional point cloud data processing. According to the invention, the point cloud classification model is trained by using the sample point cloud data, so that the trained point cloud classification model can identify the first target from the original point cloud data representing the first target; disturbing part of data points of the original point cloud data to obtain first confrontation point cloud data, and calculating classification loss of a trained point cloud classification model for identifying the first confrontation point cloud data representing the first target as a second target; obtaining a first loss representing a point cloud generalization distance and a second loss representing a grid surface generalization distance between the original point cloud data and the first confrontation point cloud data, and determining a collaborative ...

Vulnerability processing method and device, equipment and storage medium

The invention provides a vulnerability processing method and device, equipment and a storage medium. The method is applied to the technical field of vulnerability defense and comprises the steps of determining a calling tracking point of a host program in a kernel according to program debugging information and program execution data of the host program; obtaining function method calling data of the host program by utilizing the calling tracking point; feature extraction and algorithm analysis are carried out on the function method calling data, violation information is obtained according to a comparison result, and the violation information represents available vulnerabilities or other malicious operations of a host program; and generating a target virtual interception code for the violation information, and injecting the target virtual interception code into a kernel module of the host to intercept a vulnerability utilization attack on a host program. In this way, the problem that an existing ...

Wet rale data comparison method and application thereof

The invention provides a wet rale data comparison method. The wet rale data comparison method comprises the following steps: S1, taking wet rale subjected to auscultation labeling confirmation by a doctor, and segmenting the wet rale into a plurality of wet rale frequency segments; s2, segmenting a section of collected breath sound into a plurality of breath audio clips; s3, respectively segmenting all the wet rale audio segments and all the breath audio segments into eight small-segment signals, and then extracting absolute values, average values and variances of the eight small-segment signals and ratios of the absolute values, the average values and the variances to adjacent values of the average values; s4, performing Moses limit reaction test on all characteristic values of the wet rale audio clip and all characteristic values of the breath audio clip; and S5, judging whether the audio clip has the wet rale or not according to the number of the undifferentiated characteristic values ...

Buffer vulnerability recognition model training method and device and vulnerability detection method and device

The invention provides a training method of a buffer vulnerability recognition model and a vulnerability detection method and device. The method comprises the following steps: collecting an original software sample and a corresponding original label; determining buffer overflow attributes of the original software samples, and marking corresponding attribute values; performing code representation on each original software sample, and mapping the code representation into a multi-dimensional vector according to an attribute value, namely a code multi-dimensional vector; dividing the code multi-dimensional vector of each original software sample and the corresponding original label into a training set and a test set; training a buffer vulnerability recognition model by using the training set; and evaluating and/or optimizing the trained buffer vulnerability recognition model by using the test set. In this way, the buffer area vulnerability recognition model can accurately recognize software ...

Armwear face recognition method and system, electronic equipment and readable storage medium

The invention provides an armed face recognition method and system, electronic equipment and a readable storage medium, and belongs to the technical field of image processing, and the method comprises the steps: obtaining each channel matrix of a to-be-detected image of an armed face, so as to extract each channel initial dictionary matrix; randomly generating an initial pixel sparse coding matrix of each channel; iteratively updating the initial dictionary matrix of each channel to determine the dictionary matrix of each channel; orthogonal matching processing is carried out on the initial pixel sparse coding matrix of each channel; respectively carrying out maximum pooling processing on the pixel sparse coding matrix of each channel and then carrying out column series connection; carrying out pyramid pooling processing on the block link feature matrix of each channel and then carrying out column series connection; and inputting the armed face features into a trained face recognition model ...

Unmanned aerial vehicle target tracking method based on Steple

The invention provides an unmanned aerial vehicle target tracking method based on Steple, and the method comprises the steps: obtaining a first frame of image from a camera by an unmanned aerial vehicle at the beginning of a tracking task, selecting a target needing to be tracked, determining the center position and the size of the target, extracting features from the image, calculating the response, and carrying out the target tracking through employing the histogram of oriented gradient features and the size features, and obtaining a target position and scale response, calculating the target position and scale, performing confidence judgment on a tracking result, and updating the filter model. According to the method, the tracking effect is judged through dual mechanisms, so that high-confidence updating of the model is realized, errors caused by model updating are avoided, and the target tracking precision is improved. Two tracking confidence judgment mechanisms of position response ...

Random forest model training method, malicious website detection method and device

The embodiment of the invention provides a random forest model training method and device and a malicious website detection method and device, and relates to the field of data processing. The method comprises the following steps: acquiring website feature parameters of samples in a malicious website data set; carrying out normalization processing on the content of the feature parameters of each sample website and carrying out feature splicing; generating a training set according to the website feature parameters after normalization processing and feature splicing and labels corresponding to the samples; and training a random forest model by using the training set. In this way, malicious websites can be analyzed and predicted from multiple dimensions, and accurate prediction and classification of the malicious websites are achieved through the random forest model.