Millimeter wave three-dimensional holographic image denoising method

Abstract

The present invention provides a millimeter-wave three-dimensional holographic image denoising method, comprising: performing millimeter-wave scanning on a target to obtain an original three-dimensional holographic image; performing high-pass filtering on the original three-dimensional holographic image to obtain a foreground image; Linear transformation; nonlinear transformation is performed on the reflection intensity features of each point after linear transformation; K-means clustering is performed on each point of the foreground image according to each feature after transformation, and the foreground image after K-means clustering is obtained; The corresponding points in the original three-dimensional holographic image are extracted from each point of the foreground image after classification to generate a denoised image; outliers in the denoised image are removed to obtain the final image. The invention takes into account the spatial position characteristics and reflection intensity characteristics of each point, can effectively remove the noise in the millimeter wave holographic image, and reduce the amount of data; at the same time, it ensures the connectivity of the foreground image and completely maintains the spatial geometric information of the target.

Description

technical field [0001] The invention relates to the technical field of three-dimensional image processing, in particular to a method for denoising a millimeter-wave three-dimensional holographic image. Background technique [0002] Millimeter waves can penetrate smoke, clouds, dust, sea water and clothing. Compared to optical or infrared imaging, mmWave imaging reduces or even eliminates the effects of adverse weather conditions or other obstructions such as clothing and polymers; compared to cmWave or lower frequency bands, mmWave imaging provides suitable resolution. In addition, millimeter waves are non-ionizing radiation, which is safe for human exposure. Therefore, millimeter wave imaging technology has a wide range of application scenarios. [0003] According to whether there is a millimeter-wave source in the system to illuminate the field of view, millimeter-wave imaging technology can be divided into active millimeter-wave imaging and passive millimeter-wave imag...

AI Technical Summary

Problems solved by technology

Straight-through filtering is suitable for roughly extracting foreground images, with poor accuracy

Voxel grid filtering constructs a three-dimensional voxel grid according to the length, width, and height of the holographic image, calculates the barycenter point of all points in each voxel grid, and uses the barycenter point to represent all points in the voxel grid. At the same time, the geometric characteristics of the 3D image are maintained, which effectively reduces the amount of data, but it is difficult to effectively denoise

Statistical filtering assumes that the average distance between all points in the holographic image and its nearest m neighbors satisfies the Gaussian distribution, and the distance threshold can be determined according to the mean and variance. When the average distance between a point and its nearest m points is greater than the threshold, it is judged This point is an outlier and removed. This method can effectively remove the outlier in the holographic image, but the denoising effect on the area with dense noise points is poor.

Radius filtering is based on the number of adjacent points in the radius range of the spatial point, and it is set that each point has at least enough neighbors within a certain radius range, otherwise the point will be deleted; if the radius is set too large, it will be difficult for the radius filter to remove the noise near the target contour; If the radius is set too small, it is easy to remove the points in the area where the target is located, destroying the connectivity of the target, and the denoising effect on the area with dense noise points is poor

Statistical filtering and radius filtering are suitable for removing outliers in holographic images, but these methods only consider the spatial position characteristics of points, and do not consider the reflection intensity characteristics, so it is difficult to effectively remove noise near the contour of the foreground image and dense noise

Millimeter-wave 3D holographic images have many and dense noise points, and it is difficult to obtain good results by directly using statistical filtering or radius filtering to remove outliers

Images