Human hand posture estimation method and device based on human hand structure guidance in depth image

Abstract

The invention provides a human hand posture estimation method and device based on human hand structure guidance in a depth image. The method comprises the following steps: segmenting a depth image toobtain a background-eliminated human hand region; sampling the segmented depth images to obtain point clouds, and normalizing the point clouds; constructing a human hand standardized coordinate system; constructing a multi-stage point cloud network by utilizing the normalized point cloud and the constructed human hand standardized coordinate system, and performing posture-guided alignment on the point cloud in each stage; training a multi-stage point cloud network; and utilizing the trained multi-stage point cloud network to predict the joint position of the human hand. The invention providesa human hand posture estimation method guided by a human hand structure, and provides an estimation process from roughness to fineness from a palm to fingers. Practical use proves that the method hasthe advantages of high automation degree, high precision and high speed, and can meet professional or popular application requirements.

Description

technical field [0001] The invention belongs to the fields of computer vision and computer image processing, and in particular relates to a method and device for estimating human hand pose based on human hand structure guidance in depth images. Background technique [0002] Gesture operation has the advantages of no need for contact operation, small size, and more natural. The initial gesture recognition interaction mainly uses special equipment to directly obtain the spatial position of various parts of the human hand. A typical device is a data glove, which is composed of multiple sensor devices, through which the information of the hand can be input into the computer to calculate the gesture. This kind of equipment estimates the posture of the human hand more accurately, but the gloves themselves are expensive, need to be customized according to the size of the human hand, and are not convenient to carry. Later, optical markers were applied to human hand posture detecti...

AI Technical Summary

Problems solved by technology

[0006] The regression-based method models the human hand as sparse joint points. This kind of human hand does not have explicit constraints on the position of the joints. The concept of the neural network for the human hand model is driven by data, so it does not have sufficient stability.

Images