182results about How to "Solve detection difficulties" patented technology

The invention provides a method for detecting an X-ray mammary gland lesion image based on feature pyramid network under transfer learning. The method comprises the steps: 1, establishing a source domain data set and a target domain data set; 2, establishing a deformable convolution residual network layer by a deformable convolution and extended residual network module; 3, establishing a multi-scale feature extraction sub-network based on a feature pyramid structure through a feature map up-sampling and feature fusion method in combination with a deformable convolution residual network layer;4, establishing a deformable pooling sub-network sensitive to the focus position; 5, establishing a post-processing network layer to optimize a prediction result and a loss function; and 6, migratingthe training model to a small sample molybdenum target X-ray mammary gland focus detection task so as to improve the detection precision of the network model on the focus in the small sample image. According to the method, a transfer learning strategy is combined to realize focus image processing in a small sample medical image.

The invention discloses a bistatic forward-looking synthetic aperture radar ground moving target detecting method and imaging method. The detecting method includes the steps of firstly, utilizing Bulk-Deramp filtering to eliminate Doppler blurs and reducing space variations of the Doppler rate; secondly, utilizing first-order Keystone conversion to complete range migration correction of a static target and a ground moving target; then, utilizing extended orientation nonlinear frequency modulation and frequency scaling operation to equilibrate the Doppler rate of the static target; meanwhile, enabling the Doppler rate of the ground moving target to be different from the Doppler rate of the static target; finally, building a second-order fuzzy function product, completing detection on the ground moving target and estimating the Doppler rate of the ground moving target. The problem that the moving target and the static target are hard to distinguish when BFSAR is under a background with strong clutter is solved. According to the imaging method, after the detection on the ground moving target is completed, an echo of the moving target is focused by using the estimated Doppler rate of the ground moving target, and then imaging the ground moving target is completed.

The invention provides a check tool for parts in an automobile instrument panel crossbeam steering column sub-assembly with high performance-price ratio, simplicity in operation and comprehensive detection. The check tool comprises a base, a supporting mechanism and a detecting mechanism, wherein the supporting mechanism and the detecting mechanism are arranged on the base, the instrument panel crossbeam steering device sub-assembly is horizontally arranged at the top part of the supporting mechanism in the length direction, and the detecting mechanism is arranged on one side of the supporting mechanism. By adopting the structure, a three-coordinate measuring machine which is very expensive, has trivial operation and high requirements on operation staff and is used for detecting whether the parts are qualified or not in the past is abandoned, and multiple positioning benchmarking and coordinate transformation are not required, thereby overcoming the shortcomings that the measurement efficiency is low and the production requirements are difficult to meet, greatly reducing the detection cost and being particularly suitable for detecting the industrially mass-produced parts.

The embodiment of the invention discloses a target image segmentation method, device and apparatus. The method comprises the following steps of: locating an initial region of a position of a target inan image to be processed according to a condition set by the initial target region; using a color histogram and a visual saliency histogram of the initial target region as target features; Using thepre-constructed SVM model to select the target pixel points of multiple support vectors from the image to be processed, wherein the sample characteristics of each target pixel point are composed of their color histogram and visual saliency histogram; Determining a saliency region in the image to be processed based on the calculated feature distance between each target pixel point and the initialtarget region; According to the target feature and the sample feature, calculating the weight of each pixel point in the saliency region, and deleting the pixel points in the saliency region that do not satisfy the preset condition to obtain the final region as the target region, which is used for image extraction and segmentation. The present application improves the stability and accuracy of target image segmentation.

The invention brings forward a method for automatically positioning a standard tangent plane from an ultrasonic image. The method comprises the following steps: S103, constructing a radial component module to describe the anatomical structure of a target area, performing area dividing on the target area in the ultrasonic image by use of the radial component module, and obtaining a limitation area; S105, respectively detecting different anatomical structures in the limitation area by use of multiple classifiers trained by a machine learning method, and respectively obtaining detection results; S107, fusing the detection results by use of a fusion classier, and obtaining a fusion result; and S109, according to the fusion result, determining whether the ultrasonic image is the standard tangent plane. The method provided by the invention can effectively improve the detection accuracy.

The invention relates to a non-invasive fetal electrocardiogram detection device and an electrocardiogram signal data processing method. The device comprises an electrocardiogram acquisition front-enddevice and an electrocardiogram signal analysis processing software system. It includes ECG collector, analog-to-digital conversion module, wireless communication module and core processor module. The signal obtained by the ECG acquisition front end is transmitted to the core processing module through the wireless communication module, The ECG signal quality evaluation, pretreatment, maternal ECGsignal analysis and fetal ECG signal analysis are completed, and the heart rate information of mother and fetus is obtained. Combined with waveform morphology analysis, the maternal health status andfetal intrauterine status are evaluated. The device can realize the long-term, dynamic and real-time analysis and processing of fetal signals, which is of great significance to the monitoring of maternal and fetal health during pregnancy and parturition.

The invention discloses a method and a device for measuring up-conversion time resolved spectrum. A high power continuous photosemiconductor laser is driven by a high pulse pumping source to generate laser and to excite a sample to carry out up-conversion luminescence. An optical signal is split by a monochrometer, then received by a photomuitplier and converted to an electrical signal. While the monochrometer scans the wave length, a lock-in amplifier locks and amplifies the signal. The up-conversion resolved static-state spectrum is recorded and obtained. The monochrometer fixes the waver length; a digital oscilloscope monitors the pumping source and the luminescent signal to obtain the up-conversion time resolved dynamic-state spectrum. The method and the device are characterized in that the time resolved static-state spectrum and the dynamic-state spectrum in different states are measured through adjusting the pulse width, the frequency and the amplitude of the pulse pumping source; the dynamic mechanical characteristics of material up-conversion luminescence are learned; and the level structure information inside the material is analyzed.

The invention discloses a device and a method for detecting a remanufactured automobile part. The device comprises a rotary platform (1), a manipulator device (2), an optical scanning device (3) and a control display device, wherein the rotary platform (1) is used for holding an automobile part (4) to be detected; the manipulator device (2) is used for overturning the automobile part (4) to be detected; the optical scanning device (3) is used for performing image scanning on the surface of the automobile part (4) to be detected; the rotary platform (1), the manipulator device (2) and the optical scanning device (3) are electrically connected with the control display device respectively. The invention further discloses the method for detecting the remanufactured automobile part. By adopting the device and the method for detecting the remanufactured automobile part, the problem of difficulty in detecting in remanufacturing detection of the automobile part is solved, and the detection accuracy and production efficiency are increased.

The invention provides a high and low temperature thermal-humidity test chamber which comprises a test chamber body module, an electric chamber body module and a refrigerating system, wherein the electric chamber body module is arranged on the back surface of the test chamber body module; and the test chamber body module is connected with the electric chamber body module by a fast connection buckle. An inlet air ducting device and a return air ducting device are arranged on the two sides of a test room; each of the air ducting devices consists of a plurality of adjustable air ducting blocks distributed in an array manner, the air speed and the air quantity of the whole face can be adjusted by adjusting the air speed and the air quantity by every point on a single face, so that the uniformity of the air quantity and the air speed inside the test room can be ensured, and therefore, the uniformity of temperature and humidity inside the test room is ensured, and the accuracy of a testing result is ensured further.

The invention discloses an online examination invigilation system based on big data analysis. The system includes several examinee clients connected through network, a Web server, a detection unit, acentral processor, a database unit, cloud storage and one or more management clients. The online examination invigilation system based on the big data analysis is simple in design, can effectively detect online cheating behaviors such as potential proxy examination and plagiarism during the online examination, and can promptly warn examinees or terminate the examination in real time, thereby solving the problem of difficult detection of network cheating behaviors, greatly improving the efficiency of the invigilation work of the online examination and ensuring the fairness of the examination, and the system has good use and promotion value.

The invention belongs to the technical field of pulverized coal conveying safety detection devices, and discloses a pulverized coil conveying belt safety inspection robot and a use method thereof. A frame is arranged on a robot body and formed by welding square pipes and angle iron. An upper cover plate is installed on the upper side of the frame. Rope connecting devices are located in the middlesof the two sides of the upper cover plate. Protection plates are installed on the periphery of the frame. Insulating plates and first baffles are located on the lower side of the upper cover plate and arranged on the two sides of the protection plates through bolts. A first support in the frame is connected with a connecting plate through bolts. The connecting plate is installed on the frame through bolts. The rope connecting devices are connected with an external stepping motor driver through a rope. The rope traction movement is conducted through the external stepping motor driver, no driving energy source needs to be carried, and the working capacity of the robot is greatly improved; the pulverized coil conveying belt safety inspection robot is simple in structure and high in reliability; meanwhile, the robot has certain obstacle crossing capacity and is convenient to move.

The invention provides a bottled liquid foreign matter detection method and system. According to the method, the bottled liquid is inverted before entering the production line, two image acquisition devices are arranged on two different positions of the production line to perform imaging of the same area, and the abnormity of the acquired images is detected based on a sparse representation method.According to the technical scheme, the abnormity can be completely detected from the complex background so that the anti-noise performance is great and the method and the complex and high-interference environment of the actual production can be further facilitated. The used sequence image bottled liquid foreign matter detection based on sparse representation has high detection precision, great anti-noise performance and low requirement for the equipment so that the cost of the detection system can be reduced and the technical support can be provided for implementation of automation of bottledliquid foreign matter detection on the production line.

The invention provides a wireless positioning method based on time difference of arrival (TDOA).According to the method, the relative time information that a target station reaches base stations is directly calculated by utilizing the self-correlation characteristic of a circulating prefix (CP) existing in a tuning signal transmitted by a present communication system, and then the specific position of a target is calculated by a central node by utilizing time information measured by each positioning base station. The method is implemented according to the following steps: firstly, the length of a symbol and the length of the CP in a signal are determined according to the characteristics of a signal transmitted by a positioning target; secondly, self-correlation computation is performed by utilizing the self-correlation characteristic of the CP; the relative time for the target station to reach each positioning base station is obtained according to the time corresponding to the peak of the self-correlation computation; lastly, the actual position of the target is obtained by adopting the TDOA positioning algorithm according to the relative time information. Compared with the traditional method, the positioning method disclosed by the invention is easy to implement, low in computing complexity and capable of realizing real-time positioning and tracking.

The invention relates to a gate-level hardware Trojan horse detection method based on multiple characteristic parameters. The method mainly comprises two parts: characteristic parameter extraction anddata processing. Specifically, characteristic parameters such as hopping probability, correlation, controllability and observability of a gate-level netlist are extracted in the detection process, the hopping probability reflects the activity degree of circuit nodes, the correlation reflects the correlation degree between the nodes, and the controllability and observability indicate the difficulty degree of node control and observation. Then, different algorithms are designed according to different parameter characteristics to distinguish normal nodes and Trojan horse nodes. According to themethod, the gate-level netlist hardware Trojan horse detection effect can be improved; the situation of all nodes in the circuit is reflected through multiple characteristic parameters, the possibility that a design company uses an IP core provided by a third party to introduce a hardware Trojan horse maliciously modifying the circuit in the chip design stage is reduced, and therefore the method can be universally applied to gate-level hardware Trojan horse detection and has high practicability.

The invention provides a transducer specialized for polyethylene pipeline weld ultrasonic wave detection, which has a simple detection method, high sensitivity and reliable detecting result. The transducer comprises a metal housing, a wiring pole, sound absorption materials, a piezoelectric wafer and a delaying lump, wherein, the delaying lump comprises a left delaying lump and a right delaying lump, the left delaying lump is connected with the right delaying lump by a connecting rib, the piezoelectric wafer is respectively arranged on the tops of the left and the right delaying lumps, the wiring pole comprises a signal-emitting wiring pole and a signal-receiving wiring pole, wherein, the signal-emitting wiring pole and the signal-receiving wiring pole are respectively connected with the piezoelectric wafer at the two sides. The transducer is characterized in that a magnetic block is arranged in the neutral position between the left and the right delaying lumps below the connecting rib.

The invention discloses a preparation method of a hyaluronidase modified vanadium nitride paste electrode sensor. The preparation method is characterized by comprising the following steps: preparing anano VN/graphene composite paste electrode from nano VN, graphene oxide, N-ethyl-3-methylpyridine hexafluorophosphate; then, modifying the nano VN/graphene composite paste electrode with gamma-glycidyl ether oxopropyl trimethoxy silane to obtain a gamma-glycidyl ether oxopropyl trimethoxy silane modified nano VN/graphene composite paste electrode; preparing hyaluronidase stationary liquid from deionized water, bovine serum albumin, hyaluronidase and polyvinyl alcohol; finally, dropwise applying the hyaluronidase stationary liquid to the gamma-glycidyl ether oxopropyl trimethoxy silane modified nano VN/graphene composite paste electrode, and drying to obtain the hyaluronidase modified vanadium nitride paste electrode sensor. With higher electrical conductivity than a common carbon paste electrode, the electrode sensor is specific to hyaluronic acid and has high sensitivity.

The invention discloses a preparation method of a quantum dot/enzyme compounded carbon paste electrode for detecting SAM (S-Adenosylmethionine). The preparation method is characterized by comprising the following steps of doping water-soluble CdZnTe quantum dots and transmethylase in a carbon paste electrode; by taking 1-ethyl-3--methylimidazolium hexafluorophosphate as an adhesive, mixing a carbon nano-tube with graphite powder to prepare the carbon electrode, so as to obtain the quantum dot/enzyme compounded carbon paste electrode; compared with an ordinary carbon paste electrode, the quantum dot/enzyme compounded carbon paste electrode has the advantages of conducting property improved by 1 to 2 times, wide electrochemical window, simple preparation method, low cost, easy update for surface, low residual current and the like; the eelctrode can be used for quickly detecting the SAM in a sample; the method has the characteristics of high sensitivity, good selectivity, short response time and low interference, is superior to other detection methods, and is a determination method with simpleness, quickness, convenience and feasibility for the SAM. A fixed methyltransferase electrode sensor prepared by the preparation method disclosed by the invention is low in cost, simple in preparation process and good in specificity, and has the possibility of realizing automated field determination.

The invention provides a detection tool for detecting an automobile left support frame. The detection tool has the advantages that the cost performance is high, the operation is simple, and the detection is overall. The detection tool comprises a base and an operation table, wherein the operation table is arranged on the base, one side of the operation table is provided with a groove, the groove is provided with a vertical bottom surface and horizontal side surfaces, a post body which is in horizontal arrangement and is used for being inserted into a through hole arranged on the left support frame is arranged on the vertical bottom surface, the vertical bottom surface of the groove and the horizontal side surfaces of the groove are anastomotic with the outer outline of the left support frame, and the vertical bottom surface of the groove is provided with a detection hole corresponding to the left support frame to be detected. After the structure is adopted, the defects of high use price, operation complexity and high requirement on operators of a three-coordinate part qualification detection method used in the prior art are overcome, the multi-time standard positioning and the coordinate conversion are not needed, the defects that the measuring efficiency is low, and the production requirement is difficult to met are overcome, the detection cost is greatly reduced, and the detection tool is particularly suitable for detecting parts in mass production in industry.

The invention discloses a remote sensing image ship detection method and device based on an attention model, and the method comprises the steps: 1, collecting a ship remote sensing image, and carrying out the data expansion of an image data set through annotation information; step 2, preprocessing the collected remote sensing image to obtain a data set used for ship detection model training; 3, inputting the images in the training data set into a YOLOV5 attention model improved for a small-size target, and training to obtain a trained remote sensing ship detection model; 4, cutting a remote sensing image to be detected, inputting the cut remote sensing image into the trained remote sensing ship detection model, and outputting a bounding box and confidence of a ship; and mapping bounding boxes of all the cut images back to the original remote sensing image, and obtaining a final detection result after filtering repeated targets through confidence threshold filtering and non-maximum suppression. According to the invention, the problem of difficult detection caused by sparse ship distribution and too small size in the remote sensing image is solved.

The invention relates to a high spatial resolution remote sensing image building detection method based on a shadow and segmentation. The method comprises the steps that 1, shadow detection is performed on an input image; 2, the input image is segmented; 3, the adjacency relations between shadow areas and segmentation objects are calculated; 4, each shadow area is refined; 5, the adjacency relation between each refined shadow area and each segmentation object is calculated, and a building is further extracted. Shadow information of a remote sensing image is fully utilized, building targets in various shapes can be detected, the edges of the building targets can be accurately reserved, manual intervention is not needed, and the automation degree is high.