6 results about "Camera response function" patented technology

The invention discloses a semi-supervised non-contact neonatal jaundice home intelligent early warning method based on image calibration, and the method comprises the steps: building a camera response function database, and building a color constancy depth model and a skin region segmentation network; designing a strong and weak enhancement strategy by using a large amount of label-free home data, constructing a time and context comparison learning module, and performing self-supervised training on a feature encoder; a small amount of labeled hospital data and a large amount of unlabeled family data are combined, and semi-supervised fine tuning is realized through pseudo label generation and supervised comparative learning; image sequences continuously uploaded by a user are input into the model, multi-day prediction of the bilirubin level is achieved, prediction uncertainty is quantified in combination with a Bayesian method, the risk boundary is dynamically adjusted, and personalized early warning is provided. According to the method, a user does not need to use a physical colorimetric card, color measurement errors caused by model differences of mobile equipment and variability of household illumination are inhibited from a data source, and the accuracy and robustness of subsequent jaundice assessment are improved; through a training framework combining self-supervised pre-training and semi-supervised fine tuning, the understanding ability and generalization performance of the model for the sequential characteristics of jaundice are improved.

The invention discloses a cross-terminal hand back vein living body detection method based on an illumination response characteristic. The method comprises the following steps: S1, respectively collecting a multi-light-intensity vein image sequence of a target hand back and a to-be-detected hand back through a first terminal and a second terminal; s2, performing hand back contour extraction, direction adjustment and image normalization preprocessing on the two terminal images to ensure that the sizes and the orientations of the image contours are consistent; s3, extracting a hand back region of interest, and synchronously realizing region scaling and noise reduction by adopting mean filtering; s4 and S5 are combined with the camera response function and the inverse function of each terminal, illumination response offset values of pixel points in the region of interest are calculated, and a first offset feature map and a second offset feature map are generated respectively; and S6, selecting a matching template from the second offset feature map, performing sliding matching with the first offset feature map, calculating similarity through a Pearson's correlation coefficient, and comparing the similarity with a preset threshold to judge whether the to-be-detected hand back is a living body or not. According to the invention, cross-terminal hand back vein in-vivo detection with high adaptability and high accuracy is realized.

This invention discloses a sequential image fusion enhancement method based on three-axis slide table displacement sensing, belonging to the field of machine vision technology. The method includes the following steps: S1, constructing discretized state and output equations to complete the three-axis slide table motion modeling, and establishing the correlation between motion parameters and image blur kernel and pixel position offset; S2, using the inverse function of the camera response function combined with Taylor linearization, performing exposure normalization processing on multiple frames of observed images to obtain scene illumination estimates with a unified radiance scale; S3, obtaining multi-dimensional fusion weights for the image by setting a fusion weight calculation mechanism that correlates the quality features of the fused image with the slide table position; S4, completing the fusion of multiple frames of images through a weighted fusion model to obtain a high-quality single-frame fused image. Using the above method, the problems of image motion blur and uneven multi-frame imaging quality caused by three-axis slide table motion are effectively solved, achieving accurate fusion and deblurring enhancement of sequential images.

The invention discloses a degraded image target detection method based on uncertainty semantic modulation, and the method comprises the steps: generating a haze and dark light image data set according to an atmospheric scattering model and a simulation camera response function, training a proposed network, and carrying out the verification through employing a foggy day and dark light real data set. The network comprises a semantic modulation module SMM, a multi-scale feature modeling module MSFM, a cross-stage local module CSP and a YOLOX detector. The MSFM is based on wavelet transform and multi-scale space pooling, high-frequency details of the image are enhanced, and a global structure is stabilized. The SMM is based on an uncertainty theory, calculates a feature information entropy to construct an uncertainty map, guides a recovery branch to enhance a key target area, guides semantic features of a detection branch to enhance a target as spatial prior, and relieves feature conflicts between image recovery and target detection tasks. When the method is used for target detection, the target discrimination capability in a severe environment can be remarkably improved, and the omission ratio and the false detection ratio of the target are reduced.

The invention discloses a multi-exposure fusion angle resolution scattering micro-nano structure measurement method, and relates to the technical field of integrated circuit measurement. The method comprises the following steps: acquiring a multi-exposure sequence image set, and modeling a camera response function CRF; estimating an overexposed pixel value by adopting a linear extrapolation method based on a tangent line at the endpoint of a response curve, and calculating the confidence coefficient of an extrapolated pixel; when the confidence coefficient of the extrapolated pixel is lower than the basic confidence coefficient, enhancing the extrapolated pixel by using a segmented conservative estimation strategy; a correction image set and a corresponding confidence coefficient image set are obtained through iterative processing; carrying out weight distribution on the images in the corrected image set to obtain a corresponding normalized weight set; and performing multi-scale fusion according to the normalized weight set and the corrected image set to obtain a fused image. Pixel irradiance of an overexposure area is effectively reconstructed by performing pixel-level correction on an image, and high-fidelity fusion of Fourier space intensity is realized by performing multi-scale fusion on the image.

The application discloses a kind of multi-exposure fusion's angle resolution scattering micro-nano structure measurement method, it is related to integrated circuit measurement technical field.The method includes: obtaining multi-exposure sequence image set, camera response function CRF is modeled;Overexposed pixel value is estimated using linear extrapolation method based on tangent line at response curve end point, and the confidence of extrapolated pixel is calculated;When the confidence of extrapolated pixel is lower than basic confidence, segmented conservative estimation strategy is used to enhance to extrapolated pixel;Corrected image set and corresponding confidence image set are obtained by iterative processing;Weight distribution is carried out to the image in corrected image set, and corresponding normalized weight set is obtained;According to normalized weight set and corrected image set, multi-scale fusion is carried out, and fusion image is obtained.Pixel level correction is carried out to image, and the pixel irradiance of overexposure area is effectively reconstructed, and multi-scale fusion is carried out to image, and high-fidelity fusion of Fourier space intensity is realized.