Project homepage

Welcome to the documentation of this project! This is the GitBook homepage of the project, which summarizes the project's goals, functions, and quick start methods.

CA:HRSsrhJZ3NhPAgxxuq2bmm7TZ7X6KZjMT7TEdgwbpump

Project Introduction

This project focuses on stereo vision in autonomous driving (ADAS). By generating disparity maps and using columnar pixel (Stixel) estimation and segmentation technology, obstacles and roads are detected and analyzed, providing basic functions for autonomous driving environment perception.

Function Overview

• Stereo Matching Use OpenCV to match left and right binocular images and generate disparity maps.

• Stixel Estimation Extract the depth and height of each column of pixels based on disparity information.

• Stixel Segmentation Perform cluster analysis on Stixel data to identify candidate obstacle areas.

• Scene Understanding Combine the above technologies to complete the environmental perception of the scene.

Project Structure

• main.cpp: Program entry, responsible for loading data and calling main modules.

• StereoMatching: disparity map generation and post-processing.

• StereoVisionForADAS: encapsulates the process of multiple functional modules.

• StixelEstimation: extract the core logic of ground and Stixel.

• StixelSegmentation: complete Stixel clustering and object area recognition.

How to get started

Please refer to Quick Start to obtain environment dependencies and basic operations.

Data and testing

Supports testing using stereo datasets such as KITTI. For detailed instructions, please refer to Appendix B: Datasets and Test Cases.

Contribution

Any form of contribution is welcome, including code submission, problem feedback or improvement suggestions.