3results about How to "Full details" patented technology

The application discloses a brain glioma segmentation and three-dimensional visualization model training method and system based on multi-modal fusion, and comprises the following steps: inputting multi-modal medical image data into a Laplacian pyramid multi-modal fusion learning model to generate a fusion graph, then inputting the fusion graph into a U-net segmentation model comprising an encoder and a decoder to obtain a segmentation mask under the multi-modal fusion graph; and finally, based on the multi-modal fusion graph and the segmentation mask, adopting a Marching cubes (MC) algorithm to perform three-dimensional reconstruction visualization of brain regions and brain tumor region labeling. The brain glioma segmentation and three-dimensional visualization model training method and system based on multi-modal fusion provided by the application optimizes the disadvantages of insufficient information and incomplete modeling under single modal, fully utilizes the advantages of the completeness of multi-modal data, performs multi-modal fusion, segmentation and three-dimensional visualization, can more comprehensively assist surgeons in treatment, and improves the accuracy in the operation process.

This invention discloses an image stretching method and apparatus. It uses a monotonic function as a reference function, and obtains the corresponding coordinates by substituting a preset stretching inflection point into the reference function. It then calculates the difference between the reference function at the preset stretching inflection point and the gamma reference parameter (i.e., 1) to obtain the reference offset. The reference function and the reference offset are then used as the gamma parameter, causing the stretching curve to intersect the function y = x at the preset stretching inflection point. Since the reference function is monotonic, the values ​​on the stretching curve on both sides of the preset stretching inflection point are greater than 1 or less than 1, respectively. This results in the stretching curve forming a convex curve and a concave curve on both sides of the preset stretching inflection point compared to the function y = x. Therefore, when stretching an image using the stretching curve, the concavity and convexity of the stretching curve can be used to perform differentiated stretching near the inflection point, resulting in a more distinct layered X-ray image with more comprehensive detail and improved overall image contrast.

This application provides a night vision imaging enhancement method, apparatus, electronic device, and readable storage medium. The method includes: calibrating the coordinate systems of a visible light camera and a near-infrared camera into the coordinate system of a lidar to obtain calibration results; acquiring visible light images, infrared images, and 3D point cloud data of the environment in front of the vehicle using the visible light camera, near-infrared camera, and lidar; registering the visible light image and infrared image based on the calibration results and using depth information from the 3D point cloud data; extracting texture information from the registered image, separating the structure layer and texture layer of each image; classifying the separated texture layers of each image, distinguishing between noisy textures and noise-free textures, and filtering out noise in the noisy textures using a joint bilateral filtering algorithm; and fusing the noise-filtered texture layer and structure layer in the HSV color space to generate an enhanced night vision image. This method improves the safety of nighttime driving.