23 results about "Acutance" patented technology

This invention discloses an automatic calibration method for airborne lidar placement deviation based on voxel sharpness. The method includes: selecting a natural calibration site containing building facades, linear markers, and point targets; collecting point cloud and POS data via a UAV flying in multiple headings; modeling the placement deviation to be calibrated using six parameters; optimizing the parameters using a genetic algorithm; using voxel sharpness as the fitness function; voxelizing the point cloud in the region of interest; counting the number of points within each voxel and calculating the sharpness score; a higher sharpness score indicates a more pronounced clustering of point clouds in the feature region; and iteratively optimizing the method through selection, crossover, and mutation operations to output the optimal placement deviation parameters. This invention uses an ordinary basketball court as a natural calibration site, eliminating the need for artificial targets. After calibration, the point cloud sharpness is improved by more than 150%, and the positioning accuracy reaches the centimeter level.

The application provides a RAW domain low-illumination image enhancement method and system, relates to the technical field of image processing, and can effectively avoid amplification of noise in the subsequent enhancement process, reduce noise and color distortion of the RAW domain enhanced image by separating the content signal graph and the noise residual graph of the RAW domain low-illumination image. Through the content signal graph, the light rays of the camera light path are simulated and light integration is carried out, so that the physical level detail reconstruction can be carried out on the region with serious structure loss in the RAW domain low-illumination image, and the global light consistency and edge sharpness better than the traditional two-dimensional convolution network are realized. By introducing the target double-flow recurrent neural network, the optimization conflict between denoising and color restoration in the low-illumination enhancement can be solved, the progressive collaborative improvement of the image quality can be realized, and the high harmony of the final enhancement result in the three dimensions of brightness, detail and color is ensured.

The application provides an IC equipment alignment mark image enhancement method and an image enhancement device, wherein the method comprises: acquiring a process degradation mark image to be enhanced; performing real-time quality evaluation on the process degradation mark image to be enhanced by using a preset image quality evaluation algorithm; if the evaluation result is lower than a preset quality threshold, inputting the process degradation mark image to a trained cycle generative adversarial network to output an enhanced alignment mark image; the cycle generative adversarial network is trained in the following manner: training the network by using a non-paired training data set, and in the training process, optimizing the network by using a loss function comprising a structural similarity loss and an edge phase consistency loss, and training and reasoning by using a dynamic slicing strategy. By the application, the signal-to-noise ratio, edge sharpness and geometric position fidelity of the alignment mark image are improved without paired data, thereby improving the alignment accuracy.

This application provides a nighttime surveillance video enhancement method based on spatiotemporal joint and photometric adaptive mapping, belonging to the field of video image processing. The method includes: firstly, constructing a temporally correlated video frame sequence; then, decoupling the video sequence into a low-frequency illuminance feature sequence and a high-frequency reflectance feature sequence through a dual-branch decoder network; next, applying an iterative photometric adaptive mapping function to adjust the brightness of the low-frequency illuminance features, and applying deformable convolution and three-dimensional dynamic filtering to perform spatiotemporal alignment and denoising processing on the high-frequency reflectance features; finally, fusing the processed illuminance and reflectance features to reconstruct the enhanced video frame. This invention solves the problem of inter-frame temporal flicker and noise avalanche caused by brute-force brightening in traditional algorithms by completely decoupling and differentially parallelizing illuminance and reflectance components, improving details in dark areas while preserving edge sharpness, and achieving high-quality real-time enhancement of nighttime surveillance videos.

This invention relates to a method and apparatus for judging defect quality. The method includes: acquiring multiple images of an inspection object (10) comprising multiple transparent films (1,2,3,4,5,6) in the height direction at predetermined steps using an optical imaging device (22); calculating the sharpness of a portion of the images based on the brightness difference between each pixel of the multiple images and its adjacent pixels; calculating the height information of the portion of the images based on the image number of the image with the largest calculated sharpness at the same pixel position in all the images of the multiple images; obtaining three-dimensional information of all the images from the calculated height information; and judging the defect quality of the inspection object based on the three-dimensional information.

The application discloses a strip mode spaceborne SAR multi-dimensional parameter joint self-focusing imaging method. First, spaceborne SAR original echo complex data and related parameters are acquired and preprocessed, and then baseband Doppler center frequency is estimated through azimuth direction delay conjugate multiplication and two-dimensional integration. Subsequently, Doppler ambiguity is solved based on an image sharpness maximization criterion, and the absolute Doppler center frequency is obtained in combination with the baseband frequency. Finally, initial Doppler frequency modulation is acquired through contrast maximization coarse search and sub-aperture cross-correlation fine search, a multi-dimensional parameter joint self-focusing optimization model based on a minimum entropy criterion is further constructed, three variables of Doppler center frequency, Doppler frequency modulation and center slant range are cooperatively optimized by using a simplex method, and full-automatic high-precision high-resolution imaging is realized. The method can complete parameter inversion and self-focusing imaging under the condition of extreme blind processing without ephemeris, and has excellent robustness and engineering practical value while reducing the calculation cost.

The present application relates to the technical field of image recognition and defect detection, and discloses a valve sealing surface defect detection method and system based on image recognition, which comprises the following steps: acquiring an original image sequence through multi-focal image acquisition, establishing a preliminary interlayer continuity correlation of defects by using an optical flow algorithm, calculating a compensation value based on a gradient change rate to complete image compensation for pixel displacement caused by uneven reflectivity, using a convolutional neural network to reduce noise and compensate for deep image boundary sharpness attenuation, iteratively calculating defect depth values to obtain depth distribution information, tracking defect evolution paths to establish a tracking chain, generating a micro defect overall depth model, and determining the final detection result. The present application solves the core problems of reflectivity interference of the valve sealing surface, deep boundary sharpness attenuation, and insufficient precision of defect depth detection, realizes accurate identification and depth modeling of micro defects, and provides technical support for guaranteeing the sealing performance of the valve and improving the reliability of equipment operation.

The application discloses an industrial surface defect detection method and system based on sampling disturbance optimization and history reuse, and relates to the technical field of industrial visual defect detection. In view of the problems that the background of an industrial defect image is complex, the loss topography is highly non-convex, which leads to model overfitting, and the traditional sharpness perception minimization algorithm relies on double back propagation, which leads to excessive algorithm cost, a lightweight optimization method is provided. The method introduces a sampling sharpness estimation mechanism in the network parameter update stage, generates a disturbance tensor by sampling within the preset norm boundary to explore the flat minimum value, and introduces a history reuse mechanism to dynamically evaluate and reuse the historical state, so as to bypass the back propagation calculation of the current calculation graph in part of the training period. With less memory cost, the application greatly reduces the disturbance calculation cost, effectively crosses the non-convex loss valley, significantly improves the generalization accuracy of industrial surface defect detection, and has a wide application prospect.

The application relates to the field of artificial intelligence, and discloses a multi-modal fusion image repairing and enhancing method, which comprises the following steps: acquiring an RGB image and a depth map thereof; performing semantic segmentation on the RGB image to determine a missing area; performing edge-aware upsampling on the depth map; respectively extracting texture and geometric features through a double-branch encoder; realizing pixel-level dynamic alignment through a cross-modal alignment module containing a deformable convolution and a spatial attention mechanism; and generating a repaired image through a feature fusion decoder. Through the construction of the cross-modal alignment module containing the deformable convolution and the spatial attention mechanism, the application realizes pixel-level dynamic alignment of the RGB image and the depth map at the multi-scale feature level, effectively solves the feature mapping dislocation problem caused by the resolution difference of original modalities, and ensures that the depth map retains geometric boundary sharpness in the upsampling process through the edge-aware upsampling strategy, thereby avoiding the blur effect introduced by a traditional interpolation method.

The present application relates to the field of image processing, and discloses a kind of for air turbulence causes the inhibition method of image jitter, comprising the following steps: S1, multi-frame image acquisition: using general camera is continuously collected in outdoor environment the multi-frame image sequence of target object, in the present application, by pixel-level weighted fusion to multi-frame image, can make random turbulence noise mutually offset in statistical sense, while enhancing target signal, to significantly improve image signal-to-noise ratio, this method uses the weighted strategy based on image sharpness index, preferentially select the clear frame that is less affected by turbulence to participate in fusion, avoid the blurring accumulation problem caused by simple average, further, by iterative back projection algorithm is carried out super-resolution reconstruction, reconstruct high-resolution details using the complementary information of multi-frame image, compensate the resolution loss caused by turbulence, finally output single frame high-quality no turbulence image, fundamentally solve the dynamic distortion problem that traditional single frame processing method cannot overcome.

The invention relates to a projection unit (110) for projecting a test pattern (140). Sharpness characteristics of an optical test object (X) can be tested by using the test pattern (140). The projection unit (110) comprises a diffractive optical element, which is designed to produce a plurality of virtual test objects (142, 143) as the test pattern (140) in response to different excitation lights (125, 135) coupled into the projection unit (110) at the same time from one or more light sources (120, 130). In this case, different test objects (142, 143) are arranged in different focal planes and at different solid angles from the point of view of the test object (X).

This invention discloses a method for sharpening fire smoke images based on image enhancement, belonging to the field of fire image processing technology. Addressing the problems of flame highlights interfering with atmospheric light estimation and the distortion of transmittance estimation edges due to spatial non-uniform variations in smoke density in fire scenes, this invention introduces a partitioned adaptive atmospheric light estimation method based on collaborative perception of the flame-smoke region within an atmospheric scattering model framework. The image is divided into flame, dense smoke, transition, and sparse smoke regions using color-brightness joint features. After eliminating flame highlights, atmospheric light is independently estimated for each region and then weighted and fused to generate a spatially continuously varying atmospheric light distribution map. Simultaneously, a multi-scale adaptive transmittance refinement method driven by smoke density gradients is introduced. Based on the dark channel gradient, the guided filter parameters are adaptively adjusted to maintain edge sharpness at the dense smoke-sparse smoke boundary and ensure smoothness and consistency in uniform regions, suppressing halo artifacts and blocky effects.

The application provides a front wide-angle additional lens and an image pickup device, the front wide-angle additional lens sequentially comprises, from an object side to an image side, a first lens group with negative refractive power, a second lens group with positive refractive power, and a third lens group with positive refractive power, the first lens group comprises at least two lenses, and the first lens group comprises at least one lens with negative refractive power; the front wide-angle additional lens satisfies the following conditional expressions: 0.8<=|F1N| / |F1|<=1.2; 0.9<=|F1| / |F2|<=1.1. By adopting the front negative and rear positive refractive power distribution of each lens group and satisfying the specific conditional expressions, the front wide-angle additional lens has the characteristics of high sharpness and low dispersion; moreover, the front wide-angle additional lens has a small volume and can be miniaturized, and is convenient to carry.

The application relates to the technical field of image processing and fire-fighting equipment detection, and discloses a high-pressure water mist spray form image detection method and system, which comprises five steps of structured backlight imaging collection, adaptive threshold binarization and contour extraction, multi-dimensional morphological feature extraction, multi-frame time sequence statistical fusion, and classification rating and defect positioning. A high-contrast spray side image is acquired through backlight shooting, a fog cone contour is extracted through adaptive threshold binarization, four-dimensional morphological parameters of a half-cone angle, symmetry, boundary sharpness and distribution uniformity are calculated, after elimination of the influence of pulsation through multi-frame time sequence fusion, a four-level rating result is output by a classification model, and a non-standard parameter item is positioned.

This invention relates to a two-dimensional spatial laser communication beam coupling method based on sharpness-perception minimization, belonging to the field of spatial laser communication technology, and solving the technical problem that existing algorithms for improving coupling effects often fail to achieve the expected coupling results. The method includes the following steps: acquiring optical power in the neighborhood of the current spot position; performing gradient estimation based on the acquisition results; calculating the next position of the spot based on the sharpness-perception minimization algorithm; acquiring the optical power output at the adjusted spot position; if the optical power output meets the requirements, pausing coupling and continuously monitoring the optical power; otherwise, returning to the initial steps. The two-dimensional spatial laser communication beam coupling method based on sharpness-perception minimization of this invention provides more accurate gradient estimation, with a higher degree of agreement between the estimated gradient and the actual gradient; it can escape local optima and improve the ability to find the global optimum.

The application discloses a method and system for out-of-focus scanning stack diffraction imaging based on parallel iteration. The method for out-of-focus scanning stack diffraction imaging comprises the following steps: measuring a measured diffraction light field intensity generated by an actual illumination probe acting on a sample to be measured; obtaining initial probe, sample and guessed diffraction light field information; then performing weighted updating and evaluation on amplitude and phase information of the initial probe and the sample through a stack diffraction imaging algorithm until a mean square error (MSE) between the measured diffraction light field intensity and the guessed diffraction light field intensity is less than a threshold value. The iterative algorithm used in the method can process multiple groups of diffraction data in parallel, and the reconstructed probe and sample can be weighted and superimposed in the algorithm by using evaluation methods such as mean weight, image correlation and image sharpness, and the method has the advantages of fast iterative convergence speed and strong expansibility.

The application provides a front long-focus additional lens and an image pickup device, the front long-focus additional lens sequentially comprises, from an object side to an image side, a first lens group with positive refractive power, a second lens group with negative refractive power, and a third lens group with negative refractive power, the third lens group at least comprises a lens with positive refractive power and a lens with negative refractive power; the lens satisfies the following conditional expressions: 0.8<=|F1| / |F2|<=1.2; 0.8<=|F3N| / |F3P|<=1.2. By adopting the front positive and rear negative refractive power distribution of the first lens group, the second lens group and the third lens group and satisfying the specific relationship, the aberration introduced by each lens group is reduced through the multiple deflection of light rays by the lenses of the multiple lens groups, so that the front long-focus additional lens has the characteristics of high sharpness and low dispersion; moreover, the front long-focus additional lens has a small volume and can be miniaturized, and is convenient to carry.

This invention discloses an adaptive, dimensional-wise cropping differential privacy federated learning method for medical image classification. A parameter server initializes and broadcasts parameters; local medical clients in each medical institution perform local sampling and calculate gradients using a sharpness-aware minimization optimizer; adaptive dimensional cropping is performed, setting different cropping thresholds for each dimension of the gradient and adding corresponding Gaussian noise to the cropped gradients; the clients, using an adaptive learning rate mechanism, dynamically adjust the step size based on gradient fluctuations and update local parameters before uploading; the parameter server performs weighted aggregation, iterating repeatedly until all communication rounds are completed and the total privacy overhead is calculated. In practical applications, this invention effectively mitigates the negative impact of noise on model performance, significantly improving model accuracy and reducing privacy costs while ensuring sample-level differential privacy. It is suitable for collaborative modeling of sensitive image data across medical institutions.

The present application relates to a kind of nail stamping pattern edge definition compensation method, belong to image processing and stamping pattern quality control technical field.It mainly solves the problem of pattern edge blur and definition deficiency caused by image noise and direction confusion.The method is by original image gray normalization of stamping pattern, constructs multi-angle direction response convolution kernel and collects multi-scale direction features, combines local space adaptive threshold screening significant edge pixels, extracts main direction angle and statistics direction consistency, based on Shannon entropy index adaptive adjustment main, orthogonal direction gradient weight.Further, by dynamic mapping combined with neighborhood gray variance, drive direction alignment's enhanced gradient vector mapping to original pixel gray space, realize contrast re-calibration and edge definition compensation.The technology significantly improves the recognition of edge structure in complex background and pattern sharpness, provides high-precision image signal support for automatic stamping of nail sticker.

This invention discloses a medical image segmentation method based on dual-domain feature injection and adaptive edge enhancement, belonging to the field of medical image segmentation technology. This invention achieves differentiable adaptive edge extraction: the proposed ALEM abandons traditional non-differentiable and noise-sensitive operators such as Canny, and through an end-to-end learning mechanism and spatial gating technology, can accurately locate lesion boundaries in medical images filled with speckle noise. This design significantly reduces the Hausdorff distance of the segmentation results, effectively solving the problem of boundary overflow or missing values, and addressing the issue of balancing details and semantics during upsampling. Through the GDSCU module, this invention innovatively combines the stability of static interpolation with the flexibility of dynamic sampling. This collaborative mechanism ensures that when the network restores image resolution, it avoids edge blurring as in traditional interpolation and severe checkerboard artifacts as in single dynamic sampling, significantly improving the sharpness of the segmented edges.

The present application relates to a focus determination device, an autofocus system, and a substrate processing system. Whether or not focusing is achieved is determined automatically and with high precision. The focus determination device 12 includes a sharpness calculation unit 510 that calculates the sharpness of a hole image that represents a hole processed on a substrate, and a focus determination unit 520 that determines whether or not the focus of the hole image is aligned based on the sharpness calculated by the sharpness calculation unit 510.