32results about How to "Meet real-time detection requirements" patented technology

The present invention relates to an artificial sign detection method applied to augmented reality, and the specific steps include: S1: collecting a frame image, sampling the frame image roughly, using oblique grid scanning, and detecting edge pixels of the frame image; S2: based on the RANSAC algorithm , to detect the edge segments in the frame image; S3: fuse the edge segments; S4: extend and filter the edge segments; S5: detect the corner points of the quadrilateral, and construct a quadrilateral according to the corner points of the quadrilateral. The invention preprocesses frame images before operation, performs coarse grid sampling, and performs edge detection on each grid area, greatly reduces program operation time, improves detection speed, has good real-time performance, and meets real-time detection requirements. The invention adopts an edge-based detection method, firstly performs a line segment test, and then reconstructs a quadrilateral frame of a sign according to the line segment obtained by the line segment test, and has good robustness to illumination changes and occlusion conditions.

The invention provides a vehicle detection method based on multi-feature fusion. The vehicle detection method based on the multi-feature fusion comprises the steps of step 1, establishing one or a plurality of monitoring areas (ROI (region of interest)) in an input image; step 2, extracting Harris angular points and longitudinal gradient information in the ROI on the basis of a grayscale map of an image to be detected; step 3, computing a geometry size of a detection window at a certain position inside the ROI according to a circumscribed trapezoid of an obtained monitoring area; step 4, counting an angular point number of the detection window and computing the distribution symmetry of the angular points; step 5, extracting structured horizontal line information on the basis of a longitudinal gradient in the detection window; step 6, utilizing a cascaded strong classifier to judge whether a current detection window is a vehicle area, if so, reserving the current detection window, and if not, dismissing the current detection window; step 7, moving the detection widow according to a certain step size, switching to the step 3, and repeating the steps from the step 3 to the step 7; and step 8, reducing the probability of the repeated detection of the same target through detection result combination.

A method and device for extracting the center of a structured light strip for rut detection, the method comprising: obtaining an image of the center of the light strip; selecting N columns of reference columns and N columns of search columns along the normal direction of the light strip in the image of the center of the light strip , to obtain N reference column laser light bar signals and N search column laser light bar signals respectively; use the signal correlation method to correlate the N search column laser light bar signals with the N reference column laser light bar signals one by one, and obtain the maximum correlation value and the signal offset corresponding to the maximum value; according to the signal offset, determine the center position of the light bar in the middle column of the N columns of search columns, and repeatedly select different search columns until the center position of the light bar of each search column is determined . The invention well solves the disconnection problem of the laser light bar, can quickly and efficiently extract the center of the light bar, and can meet the requirements of the system for real-time detection.

The embodiment of this specification discloses a method, device, and electronic device for machine learning and privacy code determination, wherein, the machine learning method can acquire batches of sample data including positive and negative samples, the positive sample data includes privacy code files, and the negative The sample data does not contain privacy code files; based on the similarity measurement parameters of the code files, a plurality of first code files are screened out from the sample data, and the code files are screened out from a preset code library whose privacy labels are known A plurality of second code files; based on the plurality of first code files and the plurality of second code files, determine the target parameters corresponding to the sample data; combine the target parameters corresponding to the sample data and the sample data The label of is used as input to train the target model, and the target model is used to determine the similarity measurement parameters between a batch of code files to be detected and private code files.

The invention relates to a wireless electromagnetic wave detection device, which solves the problems that the current electromagnetic wave detection device is narrow in the detection frequency band, low in the detection accuracy and high in the cost. The wireless electromagnetic wave detection device comprises an antenna, a logarithm power detection device, a filtering unit and a peak value demodulation unit; the output terminal of the antenna is connected to the input of the logarithm power detection device; the output terminal of the logarithm power detection device is connected to the peak value detection unit through the filtering unit. The invention is widely applicable to the radio wave detection technology field.

The invention discloses a Newton's theorem-based method for detecting ellipses in shot object images. According to the method of the invention, in a pre-processing stage, invalid edge information is removed according to arc dimension constraints; edge gradient information is extracted, the slope of each edge point is calculated, and edges are divided into four categories according to the slope information; arcs which satisfy a condition of constituting an ellipse are aggregated according to the relative positional relationships of endpoints and Newton's theorem-based geometric constraints; theleast squares method is directly adopted to estimate the parameters of arcs belonging to the same set; and the validity of fitted ellipses is verified, ellipses that do not satisfy a verification condition are removed, and ellipses requiring repeated detection are aggregated. Since a large number of invalid edges are removed, dimensions, relative positions and Newton's theorem-based geometric constraints are adopted, the precision and speed of ellipse detection are significantly improved.

The invention discloses a system and method for visual detection of redundant objects based on deep learning in mechanical assembly. The detection system includes a workbench, an adjustable angle camera, a light source, a vision controller, etc. The vision controller processes and analyzes the images obtained from the camera to identify the objects in it, and then distinguish the assembly parts and redundant objects; the detection methods include The steps are as follows: obtain the multi-angle assembly area image, preprocess the image and input it into the trained target detection network model for feature extraction to predict the position and type of the object, then judge whether each object is redundant, and mark the redundant object position and call the police. The invention can assist manual detection of redundant objects in a specific area from multiple angles in real-time during the assembly process, has the advantages of high detection accuracy, strong real-time performance, flexible use, etc., can reduce the introduction of redundant objects in the assembly process, and enhance product reliability sex.

The invention discloses an automatic detection method for ships in an inland bridge area, which includes: 1) collecting real traffic monitoring video in the water area of ​​the inland bridge area to obtain original sampling picture data; 2) manually marking the original sampling picture data to form Mark the image; 3) use the deep convolutional neural network to learn the features of the marked image, classify the learned features, and obtain a trained deep learning model; 4) use the trained deep convolutional neural network model to The ship is detected, the position of the ship is framed by a rectangular frame, and the category of the ship is indicated. This method has good adaptability to different weather, different light conditions and dark conditions at night, can achieve a relatively high accuracy rate, and the processing speed meets the real-time requirements.