Image prediction method and device based on region dynamic screening and electronic device

Abstract

The present application relates to an image prediction method and apparatus based on region dynamic screening and an electronic device. According to an embodiment, an image prediction method may include dividing a current frame image into a plurality of sub-regions according to a predetermined standard; Determining an extended region corresponding to each sub-region, the extended region includingthe sub-region and being larger than the sub-region; Performing image recognition to determine an object of interest in the current frame image; Determining the sub-region in the extended region as asub-region of interest based on the extended region including at least a portion of the object of interest; Predicting the sub-region of interest using image information within the extended region toobtain a predicted sub-region of interest; And replacing a corresponding sub-region of interest in the current frame image with the predicted sub-region of interest to obtain a predicted frame image.By using the prediction method, the efficiency and speed of video prediction can be improved, and the prediction accuracy is not limited by the video size.

Description

technical field [0001] The present application relates to the field of image processing, and more particularly, relates to an image prediction method and device based on region dynamic screening, and related electronic equipment. Background technique [0002] Image prediction technology can be applied in many scenarios. For example, in the field of automatic driving or assisted driving, video images of the surrounding driving environment of the vehicle are often obtained through various on-board sensors such as cameras, and current and historical video images can be used to predict future conditions. Video frames, so that the assisted driving system can make certain predictions in advance and adopt corresponding driving strategies to achieve the purpose of safe driving. [0003] Traditional image prediction algorithms, including convolutional neural networks, recurrent neural networks, etc., directly process video images of the original size. In order to ensure safe driving...

AI Technical Summary

Problems solved by technology

However, as the resolution and frame rate of video images increase, the amount of calculation required for image prediction will increase sharply, so the prediction speed and accuracy will decrease instead. Even when the video resolution reaches a certain size, the traditional prediction Algorithms are difficult to run on existing hardware

Images