Local detail enhancement (LDE) is an imagery contrast enhancement method applied to visible and uncooled long wave imagery. It enhances local spatial detail through the use of a median based high/band pass filter. The generated detail channel is blended with a histogram-equalized version of the image, creating an image that contains both local detail as well as retaining some amount of global intensity. Retaining global intensity coherency allows for easier target acquisition when compared to fully local forms of contrast enhancement.

A neural network training method, an image processing method, and apparatuses thereof are provided. The neural network training method includes obtaining a first domain image and a second domain image, where the first domain image and the second domain image are unpaired images in different domains; obtaining a scaled first domain image by scaling, at an iteration, the first domain image; obtaining a training patch by cropping the scaled first domain image, where each training patch has a same number of pixels with different contents; inputting the training patch into the neural network at the iteration, and outputting an output patch; calculating a contrastive loss based on a query sub-patch and negative sub-patches selected from the training patch and a corresponding positive sub-patch selected from the output patch; and updating model parameters of the neural network based on the contrastive loss and a generative adversarial network loss.

The present disclosure relates to systems, methods, and non-transitory computer-readable media that upsample and refine segmentation masks. Indeed, in one or more implementations, a segmentation mask refinement and upsampling system upsamples a preliminary segmentation mask utilizing a patch-based refinement process to generate a patch-based refined segmentation mask. The segmentation mask refinement and upsampling system then fuses the patch-based refined segmentation mask with an upsampled version of the preliminary segmentation mask. By fusing the patch-based refined segmentation mask with the upsampled preliminary segmentation mask, the segmentation mask refinement and upsampling system maintains a global perspective and helps avoid artifacts due to the local patch-based refinement process.

The present disclosure relates to systems and methods for image processing. The method may include receiving an original image including a plurality of pixels; determining a plurality of smoothed images by applying a plurality of filtering kernels to the plurality of pixels of the original image, wherein each of the plurality of filtering kernels may be associated with a respective kernel size; determining a plurality of enhanced images by comparing the original image with the plurality of smoothed images; and generating a target image based on at least one of the plurality of enhanced images and at least one of: the original image or at least one of the plurality of smoothed images.

A method is provided for image-processing an image dataset acquired from a patient by a medical imaging apparatus, (e.g., an X-ray apparatus), wherein the image dataset includes image values associated with image points, and depicts an acquisition region of the patient containing at least one object, (e.g., a medical device), to be enhanced, which is represented by image values within an image-value interval. The method includes determining a non-linearly high-pass filtered enhancement dataset, which is confined to an image portion containing image values lying in the image-value interval. The method also includes determining a result dataset by adding to the image dataset the enhancement dataset weighted by a weighting value. The method further includes outputting the result dataset.

A method of securing biometric information may involve obtaining a digital image that contains biometric information. The method may involve identifying at least one region of the digital image that contains the biometric information and masking the biometric information. The biometric information may be a user's fingerprint and the user's fingerprint may be sufficiently masked that the masked fingerprint would not be accepted as authentic by most or all biometric authentication systems as matching the original fingerprint.

A method may include illuminating a region of a choroid of an eye of a patient with off-axis illumination from a first imaging channel, the first imaging channel being off-axis with respect to an axis of a focus of the eye. The method may also include capturing an image of the choroid, where the off-axis illumination from the first imaging channel is off-set within the first imaging channel from the image sensor. A second off-axis illumination from a second imaging channel that is off-axis from both the first imaging channel and the off-axis illumination may illuminate the same or a different region of the choroid. The captured image of the choroid may be provided to a machine learning system. Indices associated with the image to may be identified based on an output of the machine learning system.

A data processing method for rapidly suppressing background high frequency noise in a digitized image. The data processing method includes configuring a graphical processing unit to perform a first amplification process, a pixel binning process or a first interpolation process, a first low-pass filtering process, a second interpolation process, a first subtraction process, a second low-pass filtering process, a second amplification process, and a second subtraction process on an input image, so as to subtract a subtraction mask from the input image and generate a noise-suppressed output image.

Producing a privacy-protected video clip in a video management system includes retrieving a selected video clip constituting at least a portion of a stored video stream; obtaining segments of the video clip which are spaced apart in time and have a total length equal to a defined time period; combining the obtained segments of the video clip to form a background training clip; and processing the background training clip, or the separate segments, to produce a background model. The video clip is processed to produce a privacy-protected video clip, such that, for each image the video clip, the processing includes performing background subtraction, using the background model, to define foreground regions, and obscuring the defined foreground regions.

Methods, systems, and non-transitory computer readable storage media are disclosed for utilizing a plurality of neural networks in a multi-branch pipeline to generate image masks for digital images. Specifically, the disclosed system can classify a digital image as a portrait or a non-portrait image. Based on classifying a portrait image, the disclosed system can utilize separate neural networks to generate a first mask portion for a portion of the digital image including a defined boundary region and a second mask portion for a portion of the digital image including a blended boundary region. The disclosed system can generate the mask portion for the blended boundary region by utilizing a trimap generation neural network to automatically generate a trimap segmentation including the blended boundary region. The disclosed system can then merge the first mask portion and the second mask portion to generate an image mask for the digital image.