A gesture segmentation method and system based on global expectation maximum algorithm

Abstract

The invention discloses a gesture segmentation method and system based on a global expectation maximum algorithm. Including: establishing a Gaussian model of skin color; substituting the pixel values ​​of all pixels of the image to be segmented into the Gaussian model of skin color to obtain the skin color similarity of all pixels of the image to be segmented; according to the depth information of the image to be segmented and all pixels in it The similarity of skin color, to obtain a four-dimensional space model composed of points in three-dimensional space and their skin color similarity; the four-dimensional space model is divided into multiple subspaces, and in each subspace, a loss function is constructed to evaluate the fitting effect of the hypersurface, The gradient descent method is used to minimize the loss function to obtain the four-dimensional space hypersurface of the subspace, and finally obtain the maximum value of the four-dimensional space hypersurface of each subspace according to the direction of gradient ascent; the present invention can generate comparable mathematical descriptions and realize two It provides a new basis for the fusion of different modal data.

Description

technical field [0001] The invention belongs to the technical field of human-computer interaction image information processing, and more specifically, relates to a gesture segmentation method and system based on a global expectation maximum algorithm. Background technique [0002] Biometric recognition based on image information is a hot spot in computer vision research in recent years. Among them, the recognition of human biometrics has inevitably become the main research content. Further, gesture recognition research is an important research content of human biometrics. , which mainly segments, tracks and recognizes different gestures from image data, and describes and understands them. Gesture recognition technology originates from the research of digital image processing and machine learning algorithms, and gesture segmentation is the foundation and precursor of the gesture recognition research pipeline, which has a decisive impact on the final gesture recognition result...

AI Technical Summary

Problems solved by technology

[0005] However, in the problem of gesture segmentation in the traditional computer vision field, many problems still stand out, such as the segmentation is susceptible to interference, and the EM algorithm is difficult to obtain the global optimal solution. Changes such as shadows, in addition to the background, there is also the interference of skin-like objects. For example, when a human face and gestures appear in the image at the same time, it is difficult to distinguish the two through skin color, so misjudgment is likely to occur

Images