30results about How to "Narrow detection area" patented technology

The invention provides an image multifeature extraction and fusion method and the system, comprising the following steps of, extracting color features from an image to be matched, matching color with the target image, making sure the color similarity degree, if the color similarity degree exceeds the given color similarity degree threshold, entering the next step; extracting additional feature from the image to be matched, matching the auxiliary features with the target image, determining the auxiliary feature similarity degree, the auxiliary features comprise at least one of texture feature and shape feature; on the base of color similarity degree and auxiliary features similarity degree, making synthesis judgment and acquiring the synthesis similarity degree between the image to be matched and the target image, or matching the target image with the auxiliary features extracted from the matching zone of the image to be matched. The method and the system of the invention apply a cascade connected way to match from roughness to exactness, and can determine exactly the similarity zone of the target image and the image to be matched. The method and the system of the invention achieve a rapid and highly effective matching effect and save manpower and resources.

The invention relates to a vertebra positioning method based on a convolutional neural network. Vertebra positioning is an effective way for judging spinal diseases, because tissues around the vertebra have complex structures and abundant textures, diagnosis is difficult, and the efficiency is low. A vertebra positioning method based on a convolutional neural network is provided based on the aboveproblem, and therefore vertebra positioning is more accurate. The vertebra positioning method comprises following steps: A, processing vertebra CT data to increase the contrast ratio; B, carrying outsegmentation and extraction on a key area in a vertebra CT picture by means of a Faster R-CNN model; C, predicting the location of the center of mass of each segmented vertebra by means of a kernel density estimation method; and D, carrying out three-dimensional reconstruction on the processed vertebra CT picture by means of the Mimics.

The invention provides a multi-feature nighttime vehicle detection method based on machine vision. The method comprises the steps that an upper boundary line and a lower boundary line are delimited on a detected image to obtain an area needing to be detected; in the detecting area, the binary images of a red area and a white area are obtained through the color and the brightness feature of a nighttime vehicle taillight; according to the space features of the nighttime vehicle taillight, searching is conducted on the binary images of the red area and the white area to obtain candidate taillight images; then the candidate taillight images are searched for to obtain the specific information of candidate taillights; at last, the candidate taillights are matched according to the geometric feature rules of vehicle taillight pairs so that the candidate taillight pairs can be obtained. The multi-feature nighttime vehicle detection method based on machine vision is suitable for a vehicle-mounted road target detection system based on vision, and the real-time performance and the interference resistance of the target detection system are improved. The multi-feature nighttime vehicle detection method based on machine vision is suitable for the field of vehicle intelligent anti-collision early warning and the technical field of vehicle detection.

The invention discloses an improved Online Boosting and Kalman filter improvement-based TLD tracking method, and belongs to the technical field of machine vision, artificial intelligence, man-machineinteraction and target tracking. The method comprises the following steps of: (1) initialization: initializing an improved Online Boosting classifier and a P-N learning device by utilizing an initialsample set formed through selecting a target and carrying out affine transformation; (2) image tracking: selecting a feature point, tracking the feature point for twice by using an L-K optical flow method, and comparing an error between the twice tracking with a threshold value so as to obtain a tracking result; (3) image detection: obtaining a detector result through a Kalman filter, a variance classifier, the Online Boosting classifier and a KNN classifier; (4) tracking result and detection result integration: assessing confidence coefficients of the tracker result and the detector result soas to determine which module result is finally adopted; and (5) online learning: correcting the tracker result and the detector result by using the P-N learning device, and enriching the sample set.The method is capable of effectively overcoming the shielding problems, improving the speed of original methods and effectively the precision and robustness of detectors.

The invention provides a target image positioning method and device and a sight line tracking device, relates to the technical field of image detection, and is used for solving the problem that the positioning speed of a target image is low. The target image positioning method comprises the steps of extracting a plurality of image blocks from a to-be-processed image containing a target image; inputting each image block into a first neural network to obtain a probability value of the target image contained in the image block; judging the image block with the maximum probability value as a target image block; and inputting the target image block into a second neural network to obtain the position of the center of the target image.

The invention discloses a classroom student behavior detection method, relates to image processing and recognition, and mainly solves the technical problems of low accuracy, low speed and poor robustness of a current full-image detection method. The classroom student behavior detection method comprises the following steps: converting an original image of a classroom student into a grey-scale map;inputting the grey-scale map into a deep learning model to carry out head and shoulder frame detection and prone table identification, and obtaining a head and shoulder frame; determining a hand raising detection area according to the head and shoulder frame, and performing hand raising detection in the hand raising detection area. The hand raising detection area is determined according to the head and shoulder frame, and hand raising detection is performed in the hand raising detection area so that the detection input image can be greatly reduced, the detection area can be greatly reduced, the hand detection efficiency can be effectively enhanced, the hand detection accuracy can also be greatly enhanced and the robustness is high.

The present invention provides an automatic inspection system and an automatic inspection method for equipment beside a ground rail, an image acquisition module, a positioning module and a fault detection module are arranged on an inspection vehicle, the image acquisition module acquires image data of the equipment beside the ground rail, and the positioning module acquires real-time position information of the inspection vehicle; and the fault detection module identifies and analyzes the equipment beside the ground rail according to the image data and the corresponding position information during image shooting, obtains the position of the fault equipment and determines the fault type. According to the invention, full-automatic inspection of the equipment beside the ground rail is realized, the identification precision and efficiency are improved, the flexibility and expandability are good, the labor cost is reduced, and the operation safety is guaranteed.

The invention discloses a safety helmet wearing detection method, belongs to image processing and recognition, and mainly solves the technical problems of poor robustness and poor adaptability of theexisting visual detection safety helmet wearing. The detection method comprises: acquiring an original image of a construction site, and amplifying the original image by N times through image transformation to obtain an amplified image; marking safety helmet information on the amplified image; inputting the original image, the amplified image and the safety helmet information into a target detection network for learning, and performing image feature extraction to obtain face features; obtaining a face block diagram according to the face features; and determining a head area according to the face block diagram, and performing safety helmet detection in the head area. Safety helmet detection is performed in the head area so that the detection input image can be greatly reduced, the detectionarea can be greatly reduced, the detection efficiency and accuracy can be effectively enhanced, the robustness is high and the adaptability is great.

The embodiment of the invention relates to a YOLO-based massage robot acupoint detection and positioning method. The method comprises the following steps: A10, acquiring a human body original image pasted with a circular acupoint patch; A20, adopting a YOLO network model to carry out coarse positioning on the circular acupoint patch to obtain a coarsely positioned acupoint area; A30, preprocessing the coarsely positioned acupoint area, and obtaining edge points of the acupoint patch by adopting a Canny algorithm; A40, based on the edge points of the acupoint patch, determining a maximum communication area, taking the determined maximum communication area as the edge of the acupoint patch, and based on the extracted maximum communication area, acquiring accurate acupoint center information by adopting a least square method. By means of the method, the accuracy of human body acupoint recognition and positioning is achieved, and the reliability of results is improved.

The invention discloses a paper currency safety line magnetic coding detection device and a paper currency safety line magnetic coding detection method which are used for detecting the magnetic code of the safety line of a paper currency which is fed from the short edge of the paper currency. According to the paper currency safety line magnetic coding detection device, a magnetic coding detection device comprising a plurality of magnetic probes is arranged on a transfer channel; the length direction of the safety line of the paper currency is vertical to the movement direction of the paper currency; the plurality of magnetic probes are distributed from the length direction of the safety line; and the plurality of magnetic probes are arranged obliquely relative to the length direction of the safety line in a manner that the same angle is formed between each magnetic probe and the length direction of the safety line. When the safety line of the paper currency pass through the magnetic coding detecting device, the plurality of magnetic probes carry out sectional acquisition on the magnetic code of the safety line of the paper currency simultaneously; a plurality of magnetic code sections are spliced, so that the complete magnetic code of the safety line can be obtained. The above detection device is compact in structure and occupies a small space in the transfer channel; a corresponding detection area is narrow; and detection time and data processing time are short. The detection method has the advantages of few steps, high detection accuracy and high efficiency.

A method for localizing a test region of interest on an assay membrane to determine the contours of the test region and enable calibration of the location of the test region such that the same region can be localized to image an analyte of interest after an assay run. Pre-localization of the test region limits the contours of the detection area to only the test region with a reasonable margin such that background noise received by the detector can be minimized. By limiting the region of detection to a pre-localized test region improved accuracy can be achieved in flow assay membrane tests, in particular in automated analyzer systems.

The invention discloses a license plate area positioning method, device and equipment, and the method comprises the steps: collecting an image of a vehicle, and converting the image into a gray image;performing edge detection on the gray image through a first edge detection algorithm to obtain a first edge detection result, and performing edge detection on the gray image through a second edge detection algorithm to obtain a second edge detection result; performing logical OR operation on the first edge detection result and the second edge detection result to obtain a third edge detection result; filtering the third edge detection result through rectangular and connected domain region hybrid filtering; obtaining fourth edge detection results. The first edge detection algorithm and the second edge detection algorithm are combined for edge detection, the detection area is reduced, the detection area is further reduced through a rectangular and connected area hybrid filtering method, theproblem that edge detection easily causes positioning errors of the license plate area is solved, and the accuracy of the license plate positioning area is improved.

The invention relates to the technical field of automobile intelligent control, in particular to a front-back linkage automatic door control method and device, a vehicle and a storage medium. The front-back linkage automatic door control method comprises the following steps that a door opening instruction is received, and a first detection device arranged on an automobile rearview mirror is started to collect a first image of a set area range; the first image is processed, a processing result is output, and if the processing result meets a set condition, a vehicle door is opened; at least two second detection devices arranged on the vehicle door are started for real-time detection, and the first detection device is closed; the two started second detection devices are located on the front side and the rear side of the front door or the front side and the rear side of the rear door or the front side of the front door and the rear side of the rear door respectively. Opening and closing control of the vehicle door is achieved through switching of the first detection device and the second detection device, the detection range is narrowed, the precision requirement for an algorithm is lowered, and the control cost is reduced.

The invention provides a front-end detection unit device. The front-end detection unit device comprises a GIS mechanism and a detection mechanism, the GIS mechanism comprises a first shell and a GIS assembly, and the GIS assembly is arranged in the first shell; sF6 is arranged in the first shell; sF6 is arranged in the first shell; the detection mechanism comprises a second shell and a detection assembly arranged in the second shell; the detection assembly is used for monitoring the concentration of SF6 in the first shell; the first shell is communicated with the second shell; the detection assembly comprises a detection module and a driving module used for driving the detection module to move. The SF6 leakage detection device solves the problem that traditional GIS equipment cannot detect whether SF6 is leaked or not, and has the advantages of being simple in structure and low in manufacturing cost.

A system control unit sets at least one of a non-overlapped area in which an nth image and an (n+1)st image do not overlap and a low contrast area of the (n+1)st image as an in valid area that is not used for misalignment detection of the (n+1)st image. Then, the system control unit sets alignment small blocks for a motion vector search in only an area excluding the invalid area in the overlapped area in which two combining target images overlap.

The invention relates to a high-temperature dangerous area positioning method, which comprises the following steps of: installing a binocular camera and an infrared camera on a cradle head of the samedetection device, and adjusting the binocular camera and the infrared camera into parallel optical axes; the infrared camera collects an infrared image and determines a high-temperature dangerous area according to the infrared image; collecting a pitch angle and a rotation angle of the optical axis; the binocular camera collects a binocular image containing a high-temperature dangerous area as atarget area; determining a target region of interest of the binocular image according to the target region and the high-temperature dangerous region to realize target segmentation; calculating a binocular target center position according to the target segmentation result; calculating target parallax according to the center position of the binocular target; calculating a target distance according to the distance-parallax model; calculating to obtain a target coordinate value according to the pitch angle, the rotation angle and the target distance; and the physical position of the target is calculated according to the physical coordinates of the detection device and the target coordinate value to realize high-temperature dangerous area positioning. By means of the positioning method, accurate positioning of the high-temperature dangerous area can be achieved.