32results about How to "Suppress background clutter" patented technology

A method, apparatus, and article of manufacture provide the ability to generate a damage proxy map. A master coherence map and a slave coherence map, for an area prior and subsequent to (including) a damage event are obtained. The slave coherence map is registered to the master coherence map. Pixel values of the slave coherence map are modified using histogram matching to provide a first histogram of the master coherence map that exactly matches a second histogram of the slave coherence map. A coherence difference between the slave coherence map and the master coherence map is computed to produce a damage proxy map. The damage proxy map is displayed with the coherence difference displayed in a visually distinguishable manner.

The invention relates to a radar moving target long-time phase-coherent accumulation detecting method based on an RFRAF, and belongs to the technical field of radar signal processing and detection. The method comprises the following steps that (1) radar echo data are demodulated, pulsed and compressed in the distance direction, and pulse accumulation is completed; (2) parameters are initialized; (3) the RFRAT is adopted to compensate distance and Doppler migration, and then long-time interpulse phase-coherent accumulation is completed: (4) the parameters are transversely searched, an RFRAT domain detecting unit map is constructed, and constant false alarm detection is conducted; (5) target moving parameters are estimated, and moving trace points are output. The advantages of the ambiguity function and fractional order Flourier transformation are synthesized, high order phase signals of a target generated in the moving process can be flexibly matched and accumulated, the distance and Doppler migration are compensated, accumulation and gain of a radar to complex moving target signals are improved, the capacity of detecting weak moving targets in strong clutters is possessed, the moving trace points of the target can be acquired, and popularization and application value are achieved.

The invention belongs to the field of image target recognition and provides a remote sensing ship identification method based on dense feature fusion and pixel-level attention, and aims to solve the problems that a classical neural network easily identifies a plurality of dense targets as one target under a remote sensing image ship target identification task, a large number of small targets are missed, boundary frames are easy to overlap and the like. According to the main scheme, data set division is carried out on a remote sensing image data set to obtain a training set and a test set, anddata enhancement of the training set is carried out. RGB three-channel average values r < mean >, g < mean > and b < mean > of the original remote sensing image data set are calculated and the RGB three-channel values of the images are correspondingly subtracted in the expanded data set from the r < mean >, g < mean > and b < mean >; the obtained data set is input into an improved Faster RCNN network to be trained, core modules of the network are a dense feature fusion network and a pixel-level attention network, and the network outputs candidate rotation boxes and category scores of the candidate rotation boxes; and skew IOU-based rotating frame non-maximum suppression is carried out on the obtained result to obtain an identification result of the ship target in the remote sensing image.

The invention discloses a spatial variable-resolution infrared double-color imaging detection system, which comprises a double-view-field coaxial infrared double-color imaging light path system, a complex background clutter image double-color detection recognition signal processing system, an micro electro mechanical system (MEMS) scanning-based narrow-view-field imaging control system and a double-view-field image information fusion and decision system. The system mainly solves the problem that high-resolution, large-view field, real-time imaging and transmission cannot be taken into consideration at the same time in the modern optical reconnaissance, surveillance and early warning. The system has the advantages of high resolution, large view field, real-time imaging and the like, and can be used in fields of image compression, target recognition and tracking, surveillance and early warning and the like.

The invention provides an extraction method of ocean internal wave parameters of an SAR image. The method comprises steps of based on local low-rank prior, constructing a local low-rank optimization model for an image sheet including ocean internal waves; and by carrying out alternative optimized solving on the model, obtaining an optimal rotation angle shown in the description, a low-rank image sheet shown in the description and a high radiation resolution image sheet shown in the description, thereby extracting the wavelength omega of the ocean internal wave of the SAR image. According to the invention, through the constructed local low-rank optimization model, by fully use of the statistical feature of coherent spots of the SAP image and the local low-rank feature of the ocean internal wave, coherent spots of the SAR image can be well inhibited, target structure information protection and target scattering feature keeping are performed, target identification of the SAR image is effectively achieved, feature extraction can be still achieved in case of quite weak contrast ratio and damaged waveform of the ocean internal wave, and stable numerical solution can be performed on the model and is easy to achieve.

The invention discloses an infrared weak and small target detection method based on space-time joint local contrast, and relates to the field of infrared image processing and weak and small target detection. The method comprises the following steps: S1, constructing a sliding window with the size of 3*3, traversing a kth frame of image of an original sequence image, and obtaining a spatial domainlocal contrast response graph of the kth frame of image through spatial domain filtering; S2, calculating a variance value St of a continuous frame image, and obtaining a time domain local contrast response graph of the kth frame of image through time domain filtering in combination with variance value images of three adjacent frames of images; and S3, performing normalization processing on the time domain detection result and the space domain detection result, and combining the time domain detection result and the space domain detection result in a multiplicative fusion mode to obtain a space-time joint local contrast response of the kth frame of image. According to the method, spatial information and time information are fully utilized, the problems of low infrared weak and small targetdetection precision, scene robustness and the like caused by an existing method are solved, the detection performance and the low false alarm rate in infrared weak and small target detection under a complex background are improved, and the robustness of an algorithm is improved.

The invention relates to feature extraction in the field of pedestrian re-identification, in particular to a pedestrian re-identification feature extraction method based on multi-scale feature fusion.The method of the present invention comprises the following steps of improving and constructing a multi-scale feature fusion network by taking ResNet as a backbone, wherein features extracted by a layer 3 are used as shared features, two branches are followed, the two branches are respectively a Global branch and a Part branch; reducing the stride of the last layer of ResNet from 2 to 1 by the Part branch; operating the Globe branch by using a layer 4 of the original ResNet; extracting feature maps of the layer3.1 and the layer4.1, and respectively fusing a Part branch and a Global branch with feature maps of a layer 3.1 and a layer 4.1; respectively marking as layer3.1 _ p and layer4.1 _ g, and marking as layer3.1 _ p and layer4.1 _ g; subjecting the four feature vectors (layer4.4 (Part), layer4.4 (Global), layer3.1 _ p and layer4.1 _ g) to dimension reduction to form 512-dimensional features for feature fusion, and forming 2048-dimensional features for similarity measurement. According to the method, the underlying feature map can be utilized to contain more tiny detail information of the pedestrian image, so that the extracted features can distinguish similar pedestrians more easily, and the robustness of the extracted features is enhanced.

An infrared weak target detecting method based on nonnegative constraint 2D variational mode decomposition settles the problems of high difficulty for accurately estimating frequency band in which an infrared weak target exists, easy detecting result interference by noise and background clutter, and low detecting efficiency in prior art. The method belongs to the field of infrared weak target detecting technology. The method comprises the steps of acquiring an infrared image, performing infrared image preprocessing by means of bandpass filter, constructing a target function through combining a two-dimensional variational mode decomposition method and the nonnegative constraint, inputting a preprocessing result into a target function, solving according to the target function, and outputting a decomposition result, wherein the result is K nonnegative narrowband subsignals; extracting a certain narrowband subsignal which corresponds with the infrared weak target in the result, and obtaining a target subsignal; performing adaptive threshold dividing on the extracted target subsignal, determining position and size of the infrared weak target, and outputting a detecting result. The infrared weak target detecting method is used for detecting the infrared weak target.

The invention provides an infrared weak small target detection algorithm based on a multi-channel improved DoG filter. Firstly, the infrared image is normalized. Secondly, multi-channel improved DoG (IDoG) filtering is applied to the infrared image, and 8-channel filtering results are obtained. Then the maximum value of the 8-channel IDoG is processed to get the result of the multi-channel IDoG. The results of the 8-channel IDoG are compared with those of the 8-channel IDoG. After that, Gaussian enhancement is used to get the final processing result map. Finally, the position of the maximum gray value point in the result image is found, that is, the position of the infrared small target. The invention can be used for detecting small and weak targets in infrared images, and can effectivelyimprove the detection accuracy of small targets in infrared images, so as to improve the detection performance and robustness.

The invention discloses a pretreatment method for decomposed, reconstituted and strengthened infrared small targets based on singular values. The method comprises the following steps: conducting singular value decomposition (SVD) on the inputted image to obtain the singular values of the image; analyzing and estimating singular value curves of the image to obtain prediction curves of the singularvalues, subtracting the obtained prediction curves of the singular values and the original singular value curves to obtain the differential singular value curves, and conducting normalization on the differential singular value curves to obtain differential percentage curves of the singular values; and reconstituting the image after singular value decomposition (SVD) based on the differential percentage curves of the singular values to obtain the strengthened infrared small target image. By using the method of the invention, the signal-to-noise ratio of the image can be well improved, background clutters can be suppressed, and the effect of the method is superior to that of the other methods at present.

The invention discloses a point target detection method and system based on machine learning, and the method comprises the steps: determining a to-be-detected image, and dividing the to-be-detected image into a plurality of sub-images with preset sizes; filtering each subimage by adopting an MMF to obtain filtering output of each subimage, if the filtering output of each subimage is smaller than or equal to a preset threshold value, judging that the subimage is a background, otherwise, marking the subimage as a candidate subimage, and determining a filtering score of each candidate subimage; obtaining a confidence coefficient score of a target contained in each candidate sub-image through a pre-trained BP neural network classifier; and inputting the filtering score of each candidate sub-image and the confidence score of the target contained in the candidate sub-image into a Logistic regression classifier trained in advance, and judging whether each candidate sub-image contains the target or not by combining an output value of the Logistic regression classifier. The result of the MMF algorithm and the result of the BP algorithm are fused by using the Logistic regression algorithm, so that the classification result is more accurate.

The invention discloses an infrared small target detection method based on image curvature and gradient, which belongs to the field of infrared detection and comprises the following steps of: firstly, performing curvature filtering on an input infrared image to obtain a minimum curvature characteristic pattern and an average curvature characteristic pattern; secondly, setting a threshold value according to the minimum curvature feature map, and carrying out global threshold value segmentation to obtain a binary image; thirdly, performing gradient filtering on the input infrared image by combining the binarized image to obtain a minimum gradient feature map and an average gradient feature map; then, according to the minimum curvature feature map and the minimum gradient feature map, carrying out fusion to obtain a small target feature map; and finally, performing centroid point screening on the binarized image in combination with the small target feature map and the infrared image grey-scale map, and outputting all suspected target centroid points as a detection result.

The invention discloses an infrared small target detection method based on an adaptive peak gradient descent filter, and solves the problems that irregular clutters cannot be suppressed and weak target detection performance is low due to the fact that a rectangular window is used for extracting local features of an infrared image in an existing infrared weak target detection method. According to the method, the characteristic that the local pixel gray value gradient in an infrared image is reduced and the characteristic that the local contrast difference exists between an infrared weak and small target and a background are fully utilized. Irregular clutters can be inhibited; a rectangular window does not need to be set according to target size; the invention is beneficial for detecting targets with pixels smaller than 9 * 9, improves the performance of infrared weak and small target detection, and achieves the effect of early detection.

The invention provides a method and device for determining the position of a target object, computer equipment and a medium. The method comprises the following steps: acquiring continuous multi-frame images containing a target image; performing signal enhancement on a first pixel point in the target image; determining a first motion area of the target object in the continuous multi-frame image by using the position of the target object in the continuous multi-frame image; and determining the position of a second pixel point in the target image as the position of the target object in the target image. By calculating the enhancement factor of each pixel point, the local contrast of the target object is enhanced in the target image, and the accuracy of determining the position of the target object is improved.

The invention provides a long-distance target object measuring method based on a refrigeration medium wave thermal imaging rotary table. According to the method, a refrigeration medium-wave thermal imaging sensor is used for obtaining a thermal imaging image of a long-distance target object and carrying out preprocessing to obtain a high-quality image, a laser range finder is used for carrying outtarget object range finding, the height, width and relative coordinates of the target object are calculated through the thermal imaging image, turntable parameters and range finding values, and target object measurement is achieved. According to the invention, under the condition of low visibility, measurement with higher sensitivity, higher precision and smaller error can be carried out on a long-distance target object.

The invention provides a background clutter suppression method for an FMCW through-the-wall radar of an MIMO antenna, and the method comprises the steps: carrying out the reordering of original echo data according to an antenna number, carrying out the principal component analysis according to a preset frame length, estimating the background clutter, and updating a background clutter estimation result in real time through continuous iterative operation. The contribution of the current frame to a background estimation result is enhanced, so that background clutters are inhibited to the maximum extent, and the target positioning imaging effect of the radar is improved; in combination with hardware design, the background estimation frame length can be selected to achieve a better background clutter suppression effect.