49results about How to "Reduce the chance of missed detection" patented technology

The invention discloses a residue detection method, which comprises the following steps that: foreground images which are different to a background are detected, foreground pixel points are determined, and then the foreground pixel points are characterized; the persistent state that the foreground pixel points are characterized is timed, and the timed pixel points which achieve a pre-set value are extracted; a connected region formed by the extracted pixel points is detected, and is determined to be a target to be analyzed; the motion characteristics of the target is analyzed to determine whether the targets to be analyzed are the residue. The invention also discloses a residue detecting system, which comprises a target detection device, a timing treatment device, a connected region detection device and a target analysis device, wherein, the timing treatment device times the pixel points, and extracts the pixel points, the characterization state persistent time of which achieve the pre-set value, the connected region detection device detects the connected region which is formed by the extracted pixel points, so as to determine the connected region of the targets to be analyzed, and the target analysis device is used to determine whether the targets to be analyzed are the residue. The invention can precisely extract the effective pixel points of the targets, thereby precisely judging whether the detected targets are the residue.

The invention relates to a method for detecting surface defects. The method comprises: illuminating a workpiece surface to be detected through light sources at different angles when a camera photographs the surface to be detected of the workpiece every time, fusing the images into a final image and recognizing surface detects on the final image. The light sources at different angles are used in each shooting process and the images are fused so that surface detects in different images are shown in different degrees. After fusion of multiple images, the surface detects in the final image are obvious shown so that the probability of missed detection is reduced and the accuracy of the test results is improved.

The invention relates to a power grid image positioning method and an image positioning model training method. The method comprises: obtaining coordinate information of a first-level positioning object in the power grid image and a corresponding mask image through a first-level positioning network in a predetermined machine learning model; obtaining a cut image of the first-level positioning object according to the coordinate information; jointly determining coordinate information corresponding to a second-level positioning object in the cut image through a second-level positioning network according to the mask image and the cut image; and finally, positioning the object in the power grid image according to the coordinate information. Firstly, a first-level positioning object in a power grid image is positioned through a first-level positioning network in an image positioning model; and then a second-level positioning object in dependence connection with the first-level positioning object in the cut image is positioned through a second-level positioning network, so the method can achieve the positioning from the first-stage positioning object to the second-stage positioning objectunder the condition that the size of the object is smaller, and is very suitable for the object detection and recognition in a scene that the size difference of the object on a power transmission lineis larger.

The invention discloses a vehicle and traffic sign detection method, and the method comprises the following steps: obtaining an image collected by a vehicle-mounted camera, and converting the image into a grayscale image; preprocessing the grey-scale image to obtain an interested area, and performing scaling processing on the interested area for multiple times; HOG feature calculation and classification are performed on the region of interest after each time of zooming processing, so that a plurality of target region rectangular arrays containing traffic signs can be obtained; and carrying outoverlapping degree screening on all the target area rectangular arrays according to the rectangular coordinates in each target area rectangular array to obtain a traffic sign detection result. According to the detection method provided by the invention, multi-stage zooming can be carried out on the input image, HOG feature calculation and classification are carried out on each stage of zoomed image to obtain a traffic sign detection result, so that the detection precision is improved, and the probability of missed detection is reduced.

The invention provides a production-line monitoring system which includes a tester which detects a product on a production line; a connection test sensor which tests whether the product is connected with the tester; and a control device which is connected with the connection test sensor and capable of controlling the product to be kept at a first preset position. The production-line monitoring system is capable of automatically testing whether the product is connected to the tester and automatically controlling the product to be kept at the first preset position when the product is not connected with the tester so that product missed test caused by delayed operations of work personnel and product missed test caused by delayed wiring work because of other reasons are effectively prevented so that a probability that missed test of products happens is reduced and thus potential quality hazards are reduced. The invention also provides a production-line monitoring method.

The invention particularly discloses a quick measuring device and a quick measuring method on the basis of bio-impedance technologies. The quick measuring device comprises an array probe, a switching unit, a signal excitation unit, a signal acquisition unit and an analysis and processing unit. An electrode array of the array probe is connected with the signal excitation unit and the signal acquisition unit by the switching unit, the array probe comprises a substrate and the electrode array, the electrode array is embedded in the substrate, N*M micro-electrodes are arranged in an array manner to form the electrode array, and a plurality of detection regions for measuring bio-impedance in a unit of two or four micro-electrodes are formed by the electrode array. The quick measuring device and the quick measuring method have the advantages that the detection regions are increased, the repeated point selection frequencies are reduced, different depths and surfaces of to-be-measure biological tissues can be measured, accordingly, the to-be-measured biological tissues can be omni-directionally measured without dead spaces, and missing detection can be thoroughly prevented.

The invention relates to a dust detection device and a dust collector, and the dust detection device comprises the following parts: an air duct structure which is internally provided with a gas channel; an emitting part that is used for emitting first light into the gas channel; a reflection part that is used for enabling the first light to be reflected at least once in the gas channel and formingsecond light which deviates from the first light and can penetrate through the gas channel; a receiving part that is arranged in a staggered manner with the emitting part, and used for receiving thesecond light; the emitting part and the receiving part are integrated into a whole. Due to the fact that the first light is reflected at least once in the gas channel under the action of the reflection part and forms the second light deviating from the first light, dust can be sensed through the at least two light beams, and the probability of dust missing detection is reduced; as the emitting part and the receiving part are integrated, compared with the prior art, the processing cost is reduced, and the assembly efficiency is improved.

The invention discloses an immunodiagnosis kit for detecting specific IV-type dengue virus antigen, comprising a microporous reaction plate coated with monoclonal antibody 2E8A5, sample treatment fluid, monoclonal antibody 7A6A1 combined with markers, positive control, negative control, wash concentrate, color developing solution and stop solution. The kit can detect IV-type dengue virus antigen specifically, does not cross-react with other three types of serotype dengue virus antigens and detects that the supernatant sensitivity of the IV-type dengue virus is 64 times of that of the commercial kit of the same type, namely, the Pan-E Dengue Early ELISA test kit, thus greatly improving the sensitivity of detection of clinical serum samples and simultaneously reducing omission rate in clinical applications.

The invention relates to a reagent kit for distinguishing and detecting four entomoplily transmitted pathogens, in particular to a reagent kit for distinguishing and detecting dengue virus, yellow fever virus, west nile virus and chikungunya virus by utilizing multiple real-time fluorescent polymerase chain reaction technologies. The reagent kit mainly comprises reverse transcription-polymerase chain reaction (RT-PCR) reaction liquid, primer probe mixed liquid, an RT-PCR reaction enzyme system, diethylpyrocarbonate (DEPC) H2O and packing boxes for separating and packing reagent bottles or tubes. Due to the fact the reagent kit can accurately distinguish the dengue virus, the yellow fever virus, the west nile virus and the chikungunya virus, the reagent kit can be widely applied to the multiple fields of differential diagnosis, inspection & quarantine, prevention and control of epidemic situation and the like of entomoplily transmitted virus.