Method for converting texture image into tactile signal based on deep learning

Abstract

The invention relates to a method for converting a texture image into a tactile signal based on deep learning, and belongs to the technical field of artificial intelligence and signal processing. Themethod comprises the steps of firstly, learning to train texture image data to obtain feature information of an image, so as to classify the various types of textures; using a short-time Fourier algorithm to convert a triaxial acceleration signal of frictional vibration of a material surface into a frequency spectrum image, and then conducting training to obtain a frequency spectrum generator; combining the classification information with the frequency spectrum generator to automatically generate a frequency spectrum of a texture image, converting the frequency spectrum into tactile signals ofdifferent types of images, so as to realize conversion of different texture images to tactile signals. A result is transmitted to a palm through a tactile feedback device connected to the inside of amouse, and an area where the mouse pointer is located is a material area to be tested, so that feedback of the material property of an measured object is achieved in real time by sliding of the mouse. The conversion result has high similarity to the real touch of an image texture, the application scenes are rich, and the method has high practical value.

Description

technical field [0001] The invention relates to a method for converting texture images into tactile signals based on deep learning, and belongs to the technical fields of artificial intelligence and signal processing. Background technique [0002] With the development of global industrialization and the rise of artificial intelligence, object material recognition has been widely used in many industrial fields such as e-commerce, machinery manufacturing and intelligent robots. The current material recognition is usually based on the texture image of the surface of the object to identify the material of the object in the image (such as wood, glass, plastic, steel and fiber, etc.). However, texture image-based material recognition is easily affected by the shooting environment, and large intra-class appearance differences and small inter-class appearance differences usually lead to weakened distinguishability and robustness of texture features. In addition, the texture image c...

AI Technical Summary

Problems solved by technology

Due to the complex mapping between the input and output of tactile signals and the limitations of the experimental tool level, it is difficult to take the state of the tool and the state of the surface of the texture image as input at the same time. At present, there is no experimental model that can give good feedback on other Corresponding tactile signal

In addition, the evaluation of the quality of the tactile generation results is still based on the judgment of artificial subjective feelings, and there is no objective and quantitative evaluation system, which makes the research results not have a good evaluation standard

Images