37results about How to "Reduce the false alarm rate of detection" patented technology

The invention discloses an adaptive fault detection method for an airplane rotation actuator driving device based on deep learning. According to the invention, adaptive fault detection is carried out on the airplane rotation actuator driving device based on a sparse Dropout automatic coder and a noise reduction automatic coder and deep learning of Logistic regression, feature self-learning of original data is realized through using the Dropout automatic coder in a first layer and a layered noise reduction automatic coder model in a second layer and a third layer by adopting a multi-layer neural network based deep learning autonomous cognitive method, data features acquired by learning are inputted to a Logistic regression model so as to judge an operating state of the rotation actuator driving device, a threshold is enabled to change along with different inputs and different states of the system through additionally arranging an adaptive threshold fault observer, and a residual error caused by non faults is eliminated. The method disclosed by the invention can be effectively applied to fault diagnosis of the airplane rotation actuator driving device.

The invention discloses an unmanned aerial vehicle video small-object detecting method based on super-resolution reconstruction. The unmanned aerial vehicle video small-object detecting method comprises the steps of selecting one input image as a reference frame, and selecting three consecutive image frames for performing sub-pixel displacement estimation on the reference frame; then placing the displacement estimation result of the four image frames into a high-resolution image grid; and estimating pixels which are lost in the high-resolution image grid, thereby obtaining an objective image with a relatively high resolution. Afterwards, an objective template is extracted from the objective image, and the characteristic of the objective image is calculated. Then the reconstructed objective image is divided for obtaining a plurality of objective area blocks. Characteristic extraction and characteristic identification are successively performed on all objective area blocks, thereby finishing preliminary detection for the object. Afterwards, false object elimination is performed, thereby obtaining a final detection result.

The invention discloses an ultraviolet image based water surface oil spill detection system and method and relates to the field of signal processing. The ultraviolet image based water surface oil spill detection system comprises an ultraviolet imaging module, a video decoding buffer module, an image processing module and an alarm module which are connected in sequence. The ultraviolet imaging module comprises an ultraviolet light source, an ultraviolet bandpass filter and a camera. The video decoding buffer module comprises a video decoding chip and a synchronous dynamic random access memory. The camera is connected to the video decoding chip and the video decoding chip is connected to the synchronous dynamic random access memory. The image processing module comprises a DSP (Digital Signal Processor) chip and a program loading chip. The video decoding chip, the synchronous dynamic random access memory and the program loading chip are all connected to the DSP chip. The alarm module comprises a relay, a sound and light alarm and a wireless data transmitter. The ultraviolet image based water surface oil spill detection system and method is good in water surface oil spill detection effect and high in anti-interference capability.

The invention provides a course error correcting method. The course error correcting method comprises the following steps: (a) collecting magnetic field measurement data, gyroscope measurement data and accelerometer measurement data; (b) calculating detection statistics; (c) determining whether a magnetic field is in a stable state or not in a corresponding time interval between an s step and a (s+m-1) step of a person; and (d) if the magnetic field is in the stable state, adopting a magnetic field solved course as an observation amount, estimating a course angle error by utilizing Kalman filtering, and correcting the course of the person by utilizing the course angle error. The invention also provides a course error correcting apparatus. The course error correcting apparatus comprises an accelerometer, a gyroscope, a magnetometer, a signal acquisition unit and a signal processing unit. The invention also provides a magnetic field detection method and apparatus.

The invention discloses a local gradient trilateral-based multi-scale infrared weak and small target detection method for a graph domain. The method comprises the steps of converting an infrared image containing a weak and small target into a local gradient trilateral graph signal taking a node and an edge weight relationship as an expression mode; secondly, performing multi-scale decomposition on the local gradient trilateral graph signal according to multi-scale transform of a graph Laplacian matrix to obtain low and high-frequency sub-bands of the graph signal under different scales; thirdly, performing local weighting on the high-frequency sub-band of each scale according to the edge weight relationship of the graph signal, taking a mid-value as a new center node coefficient, and performing multiplicative fusion on the high-frequency sub-bands subjected to the local weighting; and finally, performing adaptive threshold segmentation on the high-frequency sub-bands subjected to the multiplicative fusion, determining a target space position, and outputting a detection result.

The invention provides a multi-camera complex scene self-adaptive vehicle collision early-warning device and method. The multi-camera complex scene self-adaptive vehicle collision early-warning deviceis characterized in that a multi-camera sensing module includes three fixed focus cameras whose focal lengths are sequentially reduced, and the multi-camera sensing module transmits the detected front image information to a data processing and calculation module; the data processing and calculation module performs vehicle detection, vehicle tracking, and front and rear distance calculation on theimage information according to a set built-in collision warning method, and generates collision early-warning information; and a collision warning module receives the collision early-warning information, and generates warning information to a driver for warning prompt. As the three fixed focus cameras of the multi-camera complex scene self-adaptive vehicle collision early-warning device cover thewide detection range, and also avoid the problems of high cost and large size caused by using a zoom camera. The multi-camera complex scene self-adaptive vehicle collision early-warning method performs scene classification on classifiers to enable the cascade classifiers to be small in volume, high in detection efficiency, and more highlighted in vehicle features in the corresponding scene, so that the complex scene adaptive ability of the system is high.

The invention relates to a target detection method, device and equipment, computer equipment and a storage medium. The method comprises the following steps: obtaining a target image, and carrying outvariance weighted local entropy processing on the target image to obtain a variance weighted local entropy image of the target image; Further performing binarization processing and clustering analysison the variance weighted local entropy image to obtain a processed variance weighted local entropy image; Obtaining a fusion image according to the processed variance weighted local entropy image andthe local contrast enhancement image of the target image; And analyzing the fused image to determine a target in the target image. By the adoption of the method, background interferents in the improved local contrast image are restrained, the accurate position of the infrared weak small target is obtained, and the detection false alarm rate is greatly reduced.

The invention relates to a crossing bridge and offshore ship joint detection method in an onboard remote sensing image. The crossing bridge and offshore ship joint detection method comprises the steps of firstly, carrying out otsu self-adaptation threshold segmentation on the onboard remote sensing image, then sequentially carrying out morphology closed operation and morphology open operation on a segmentation result image, obtaining a water body district image, then carrying out binaryzation operation on the water body district image, obtaining a water body district binary image, separating target candidate districts and water body background districts, calculating the area average value of the target candidate districts, checking whether the water body background districts exist around the target candidate district or not, if the value of the area of one target candidate area is larger than the average value and is surrounded by the water body background districts, determining that the target is an offshore ship, and if the value of the area of one target candidate district is smaller than the average value and only two opposite sides of the target candidate district are the water body background districts, determining that the target is a crossing bridge. The crossing bridge and offshore ship joint detection method is high in calculating efficiency, real-time processing can be easily carried out on an onboard computer, and the detection false alarm rate is low.

The invention discloses a synthetic aperture radar (SAR) target detection false alarm elimination method based on priori scene knowledge, wherein the SAR target detection false alarm elimination method mainly settles a problem of large number of SAR image target detection false alarms in a complex scene. The SAR target detection false alarm elimination method comprises the following steps of (1) inputting an original SAR image, performing down-sampling twice and mean filtering on the original SAR image; (2) performing area growth on the SAR image after preprocessing, eliminating an object-sized communicating area which are not grown out, and using the rest non-grown-out communicating area as the area to be classified; (3) selecting a training sample, performing classification on the area to be classified, and obtaining an area in which an object impossibly exists; and (4) eliminating the false alarm that is detected in the area in which the object impossibly exists, and obtaining a result image. The SAR target detection false alarm elimination method can effectively reduce detection of false alarm in the complex scene and can be used for afterprocessing in detection of the SAR image target.

The invention relates to a remote sensing image riverway target detection method and device based on feature fusion, and a computer readable storage medium. The method comprises steps of a remote sensing image containing a riverway being collected and inputted; performing enhancement preprocessing on the remote sensing image by using a multi-scale contrast improvement method; shear wave transformation being carried out on the image after enhancement preprocessing to obtain variance maximum shear wave feature maps of different scales and directions; extracting an SUSAN edge feature map from theimage after enhancement preprocessing by using a similarity comparison method; calculating information entropies of the shear wave feature map and the SUSAN edge feature map, and obtaining a featurefusion map by using an information entropy weighting method; and performing threshold segmentation and morphological operation on the feature fusion image to obtain a final river target detection result image. According to the method, the capability of river channel detection under the remote sensing image is improved, and problems of low precision and high false alarm rate of traditional river channel detection are avoided.

The invention relates to a low-altitude sound target comprehensive identification method and device based on a multi-dimensional feature space. Signals are collected through a multi-channel sound sensor array; the acquired signals are analyzed and calculated to obtain a multi-dimensional feature vector including time-frequency features, spatial features and harmonic features; and the multi-dimensional feature vector is inputted into a target classification model to carry out target identification. The method can effectively improve the target identification rate.