A surround view stitching system based on an RK3568 chip and an acceleration method thereof, computer readable medium

By utilizing the RGA hardware module and Mali GPU on the RK3568 chip for image format conversion and parallel computing, a pipelined task architecture was designed to solve the real-time operation problem of the panoramic image function on a low-cost embedded chip, achieving real-time stitching and cost reduction.

CN122415320APending Publication Date: 2026-07-17CHANGZHOU RES INST OF CHINA COAL TECH & ENG GRP +1

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
CHANGZHOU RES INST OF CHINA COAL TECH & ENG GRP
Filing Date
2026-03-17
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing surround view imaging capabilities are difficult to implement in real time on low-cost, low-power embedded chips, forcing manufacturers to choose higher-cost chip solutions.

Method used

The RK3568 chip is used, and its internal RGA hardware module replaces the CPU for image format conversion. Combined with Mali GPU for parallel computing, a pipelined task architecture is designed to optimize the image stitching process and reduce memory copy overhead.

Benefits of technology

This technology enables real-time output of stitched images on the RK3568 chip, reducing CPU load, improving system stability and responsiveness, and significantly reducing material costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122415320A_ABST
    Figure CN122415320A_ABST
Patent Text Reader

Abstract

The present application relates to the technical fields of computer vision and intelligent driving system, and particularly relates to a surround view splicing system based on RK3568 chip and an acceleration method thereof, which acquires images in real time through multiple cameras, stores the acquired images in NV12 format into a frame queue, and acquires multiple synchronous frames; converts the image data of the multiple synchronous frames from NV12 format to RGB format in parallel, and stores them into a preset double buffer area; allocates GPU threads for each target pixel of the spliced result image using Mali GPU, performs table mapping operation, splices the multiple RGB images into a surround view, and stores the spliced result into a preset double buffer area of spliced images; and reads the latest spliced result from the double buffer area of spliced images and outputs it after the GPU completes the splicing of the current frame. The present application fully utilizes the heterogeneous computing resources inside the RK3568 chip, and runs the surround view image splicing software on this low-cost chip in real time, which reduces the overall material cost compared with other technical solutions.
Need to check novelty before this filing date? Find Prior Art