Example implementations relate to joint depth prediction from dual cameras and dual pixels. An example method may involve obtaining a first set of depth information representing a scene from a first source and a second set of depth information representing the scene from a second source. The method may further involve determining, using a neural network, a joint depth map that conveys respective depths for elements in the scene. The neural network may determine the joint depth map based on a combination of the first set of depth information and the second set of depth information. In addition, the method may involve modifying an image representing the scene based on the joint depth map. For example, background portions of the image may be partially blurred based on the joint depth map.

An image processing method, an image processing device, an electronic device, and a non-transitory computer readable storage medium are provided. The image processing method includes: obtaining an input image which includes M character rows; performing global correction processing on the input image to obtain an intermediate corrected image; determining the M character row lower boundaries corresponding to the M character rows according to the intermediate corrected image; and determining the local adjustment reference line and M retention coefficient groups based on the intermediate corrected image and the M character row lower boundaries; determining M local adjustment offset groups corresponding to the M character rows according to the M character row lower boundaries, the local adjustment reference line and the M retention coefficient groups; performing local adjustment on the M character rows in the intermediate corrected image according to the M local adjustment offset groups to obtain the target corrected image.

The present invention relates to a method for segmenting a digital image, for example to accurately segment cerebral vasculature on MRI-TOF images of a brain. The method first uses a model that imitates the perception of luminance contrasts by a human observer to accentuate a contrast between structures of interest, such as cerebral vasculature, and the image background. Then, the image is thresholded using an adaptive threshold. This enhanced segmentation method can be used to process digital images before launching further machine-implemented characterizations of the structures of interest, such as detecting and characterizing bifurcations of the cerebral vasculature for intra-cranial aneurysm prediction.

A method with image processing includes: setting an offset window for an offset pattern of a kernel offset and an offset parameter for an application intensity of the kernel offset; determining an output kernel by applying the kernel offset to an input kernel based on the offset window and the offset parameter; and adjusting contrast of a degraded image using the output kernel.

An automatic abnormal cell recognition method, the method including: 1) scanning a slide using a digital pathological scanner and obtaining a cytological slide image; 2) obtaining a set of centroid coordinates of all nuclei that is denoted as CentroidOfNucleus by automatically localizing nuclei of all cells in the cytological slide image using a feature fusion based localizing method; 3) obtaining a set of cell square region of interest (ROI) images that are denoted as ROI_images; 4) grouping all cell images in the ROI_images into different groups based on sampling without replacement, where each group contains ROW×COLUMN cell images with preset ROW and COLUMN parameters; obtaining a set of splice images; and 5) classifying all cell images in the splice image simultaneously by using the splice image as an input of a trained deep neural network; and recognizing cells classified as abnormal categories.

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 lighting emulation system accesses an image that is captured by an imaging device and that depicts an operational scene illuminated by close-range light. The lighting emulation system accesses a depth map of the operational scene that includes depth data corresponding to each pixel in the image. Based on the depth map, the lighting emulation system determines a far-range lighting coefficient for each pixel in the image. Specifically, the far-range lighting coefficient for each respective pixel is determined based on the corresponding depth data included in the depth map for that respective pixel. Based on the image and the far-range lighting coefficient for each pixel in the image, the lighting emulation system generates a processed image depicting the operational scene as illuminated by far-range lighting, and provides the processed image for presentation on a display screen.

In a real-time traceability method of a width of a defect based on divide-and-conquer provided by the present invention, through the calibration transfer function, the multidimensional eigenvector analysis technology based on the electromagnetic field simulation database of defect scattered dark-field imaging and the adaptive threshold segmentation method, the real-time traceability of the width of the defect greater than and close to the diffraction limit of the system is performed, respectively. The extreme random tree regression model is trained by multidimensional eigenvector analysis technology based on the multidimensional eigenvectors in the electromagnetic field simulation database of the defect scattered dark-field imaging. The present invention solves the problems that the width of the defect in defect detection is difficult to be accurately measured in real time, and the conventional image processing algorithm is difficult to accurately identify the width of the defect close to the diffraction limit of the system.

A method for automatic registration of dental image data comprises receiving three-dimensional intraoral surface scan data of a dentition of a patient, receiving three-dimensional volumetric scan data of the dentition, generating a first set of descriptors characterizing tooth crown surfaces of the dentition from the intraoral surface scan data, generating a second set of descriptors characterizing tooth crown surfaces of the dentition from the volumetric scan data, determining a best-fit registration transform based on the first and second sets of descriptors, and aligning the intraoral surface scan data and the volumetric scan data based on the best-fit registration transform.

Methods and systems for reducing banding artifacts when displaying images are described. Identified image bands are filtered using an adaptive sparse finite response filter, where the tap-distance in the sparse filter is adapted according to an estimated width of each image band. Image debanding may be performed across multiple pixel orientations, such as rows, columns, a 45-degree angle, or a −45-degree angle. Given a threshold to decide whether sparse filtering needs to be performed or not, an iterative debanding process is also proposed.