Improved depth image enhancing algorithm based on Gaussian mixed model

Abstract

The invention relates to an improved depth image enhancing algorithm based on a Gaussian mixed model. The improved depth image enhancing algorithm comprises the following steps: performing Gaussian modeling on a scene, obtaining a background matched model, acquiring a current scene image by a kinect sensor, performing hole repairing, and obtaining a hole-free background depth image; acquiring a texture map of a random dynamic scene and a corresponding original depth image by the kinect sensor, and aligning and cutting the texture map and the depth image; performing foreground and background separation on the original depth image by the Gaussian mixed model, and obtaining an original foreground image and an original background image; respectively performing hole filling on the original foreground image and the original background image; performing median filtering on the filled image. Compared with the prior art, the improved depth image enhancing algorithm overcomes the defects that the conventional method is complex, the amount of calculation is large, and image edges are blurred, and has the advantages of being simple, convenient, quick, economic and the like.

Description

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine

Login to View More

AI Technical Summary

Benefits of technology

This new invention simplifies the way we look at images while still maintaining their original quality. It achieves this by creating an algorithm that accurately matches two grayscale versions (the ground truth) from one another's reference frame. By doing these calculations over multiple frames, it becomes possible to extract important features like edge detail or texture details on both sides of each pixel within the scene being captured. These extracted characteristics help us make better understanding how objects appear when displayed together more clearly than if they were separately shown alone.

Problems solved by technology

Technological Problem addressed in this patents describes how current deep camera techniques for obtaining depth images suffer issues like instability or lack accuracy when used at different levels of detail (DOI). Additionally, existing algorithms may result in blurry edges caused by imperfections in the scene being captured during imagery capture. Current solutions involve post-processed depth images after capturing them, leading to slow down overall performance.

Images