Image Processing Method for AR Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing methods for augmented reality in electronic devices face challenges with high network bandwidth consumption when transmitting continuous image frames for object recognition, and they also burden the device with continuous image processing tasks.
Innovation Solution
An image processing method that detects motion between the electronic device and objects, allowing for the recognition of objects by determining if the same object exists in continuous frames, thereby reducing the need for continuous image transmission and processing by using a feature point extraction and compression module to transmit only necessary data to a server for comparison.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If continuous image frames are transmitted to a server for object recognition, then object recognition accuracy is improved, but network bandwidth consumption increases
Solution Approach 1:
The patent extracts only the essential feature points from continuous image frames instead of transmitting entire frames. The feature point extraction module identifies and transmits only the critical visual features necessary for object recognition, significantly reducing network bandwidth consumption while maintaining recognition accuracy.
Solution Approach 2:
The patent applies partial action by transmitting only a subset of image data (feature points) rather than complete image frames. This selective transmission approach provides sufficient information for object recognition without the excessive bandwidth consumption of full-frame transmission.
2Measurement precision
If the electronic device directly performs continuous image frame processing, then object recognition capability is improved, but device processing burden increases
Solution Approach 1:
The patent segments the image processing task into two parts: local feature point extraction performed by the electronic device, and comprehensive object recognition performed by the server. This division reduces the processing burden on the device while maintaining recognition capability.
Solution Approach 2:
The feature point extraction module acts as an intermediary that prepares simplified image data before transmission. By preprocessing images to extract only essential features, the device reduces its processing burden while ensuring the server receives sufficient information for accurate recognition.
3Loss of energy
If feature point data is transmitted to the server, then network bandwidth usage is reduced, but data transmission requirements change
Solution Approach 1:
The patent changes the parameters of transmitted data from complete image frames to extracted feature points. This parameter transformation significantly reduces data volume and network bandwidth usage, though it requires implementing feature point extraction functionality in the device.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image processing method and an electronic device implementing the method are provided. The method includes the electronic device sequentially receiving an image captured by a camera and determining whether a frame of the received image satisfies a predetermined condition. If the predetermined condition is satisfied, the electronic device recognizes an object in the frame. Then the electronic device tracks the object in the frame through tracking data created based on a feature extracted from the recognized object.