图像处理和计算机视觉
R2026b借助 MathWorks® 的图像处理和计算机视觉产品，您可以执行端到端的处理工作流，从数据采集和预处理，到增强和分析，一直到部署到嵌入式视觉系统。
这些产品支持面向图像、视频、点云、激光雷达和高光谱数据的各种工作流。使用这些产品，您可以：
使用 App 以交互方式可视化、探查、标注和处理数据。
用算法增强和分析数据。
使用深度学习执行语义分割、目标检测、分类和图像到图像的转换。
与硬件对接，用于图像采集、算法加速、桌面原型构建和嵌入式视觉系统部署。
适用产品: 图像处理和计算机视觉
Image Processing Toolbox
执行图像处理、可视化和分析
Computer Vision Toolbox
设计和测试计算机视觉系统
Visual Inspection Toolbox
Design and test visual inspection and analysis applications
Point Cloud Toolbox
Design, analyze, and test point cloud processing systems
Medical Imaging Toolbox
Visualize, register, segment, and label 2D and 3D medical images
Vision HDL Toolbox
Design image processing, video, and computer vision systems for FPGAs and ASICs
Image Acquisition Toolbox
Acquire images and video from industry-standard hardware
主题
标注和预处理数据
- 选择用于标注真值数据的 App (Computer Vision Toolbox)
决定使用哪个 App 来标注真实值数据：图像标注器、视频标注器、真实值标注器、激光雷达标注器、信号标注器或医学图像标注器。 - Get Started with Image Preprocessing and Augmentation for Deep Learning (Image Processing Toolbox)
Preprocess data for deep learning applications with deterministic operations such as resizing, or augment training data with randomized operations such as random cropping. - Get Started with Vision-Language Models (Computer Vision Toolbox)
Use vision-language models for multimodal tasks such as image captioning, zero-shot classification, and image search. - Choose an Image Registration Technique (Image Processing Toolbox)
Register images using the Registration Estimator app, or select from techniques including automated feature matching, intensity-based registration, or control point registration.
删除和跟踪目标
- Get Started with Object Detection Using Deep Learning (Computer Vision Toolbox)
Perform object detection using deep learning neural networks such as YOLOX, YOLO v4, RTMDet, and SSD. - Object Detection in Point Clouds Using Deep Learning (Point Cloud Toolbox)
Detect 3-D bounding boxes for objects in a point cloud.
分割图像和点云
- Get Started with Image Segmentation (Image Processing Toolbox)
Get started with tools for image segmentation, including Segment Anything Model, classical segmentation techniques, and deep learning-based semantic and instance segmentation. - Segment Objects Using Segment Anything Model (SAM) in Image Segmenter (Image Processing Toolbox)
Interactively segment objects or automatically segment the entire image using the Segment Anything Model (SAM) in the Image Segmenter app. (自 R2024b 起) - Semantic Segmentation in Point Clouds Using Deep Learning (Point Cloud Toolbox)
Assign class labels to each point inside a point cloud using deep learning.
筛选和分析图像
- Get Started with Image Filtering (Image Processing Toolbox)
Get started with techniques for image filtering. - Calculate Properties of Image Regions Using Image Region Analyzer (Image Processing Toolbox)
Calculate the properties of regions in binary images and identify the region with the largest area by using the Image Region Analyzer app. - Get Started with Visual Inspection (Visual Inspection Toolbox)
Learn about Visual Inspection Toolbox™ capabilities for automated inspection, defect detection, metrology, and object counting in industrial applications.
三维视觉和视觉 SLAM
- Choose SLAM Workflow Based on Sensor Data (Computer Vision Toolbox)
Choose the right simultaneous localization and mapping (SLAM) workflow and find topics, examples, and supported features. - What Is Structure from Motion? (Computer Vision Toolbox)
Estimate three-dimensional structures from two-dimensional image sequences. - What Is Multi-Sensor Calibration? (Computer Vision Toolbox)
Estimate the geometric relationships between multiple sensors mounted on the same platform.
设计光学系统和采集图像数据
- Get Started with Optical Design and Simulation (Image Processing Toolbox)
Use optical design and analysis tools to simulate and tune optical system performance for optics applications. - Get Started with Image Acquisition Explorer (Image Acquisition Toolbox)
Use the Image Acquisition Explorer to preview, configure, acquire, and save image data.
在硬件上部署
- Deploy Visual Inspection Code, Models, and Applications (Visual Inspection Toolbox)
Deploy visual inspection code, models, and applications to production environments using ONNX export, code generation, or standalone application packaging. - GPU Code Generation Workflow (GPU Coder)
Design, implement, and verify generated CUDA MEX for acceleration and standalone CUDA code for deployment. - Integrate YOLO v2 Vehicle Detector System on SoC (Vision HDL Toolbox)
Simulate a YOLO v2 vehicle detection algorithm that contains FPGA and ARM sections for deployment to an SoC device.
精选示例
调整图像大小
通过指定乘法缩放因子或指定所需的图像输出大小来更改图像的大小。
Detect Nuclei in Large Whole Slide Images Using Cellpose
Detect cell nuclei in whole slide images (WSIs) of tissue stained using hematoxylin and eosin (H&E) by using Cellpose.
Automatically Detect and Recognize Text Using Pretrained CRAFT Network and OCR
Perform text recognition by using a deep learning based text detector and OCR.
Detect, Classify, and Track Vehicles Using Lidar
Detect, classify, and track vehicles by using lidar point cloud data captured by a lidar sensor mounted on an ego vehicle.
Localize Screen Manufacturing Defects Using Student-Teacher Anomaly Detector
Train a Student-Teacher anomaly detector on normal screen images, then localize manufacturing defects by comparing student and teacher feature outputs.
Performant and Deployable Monocular Visual SLAM
Use visual inputs from a camera to perform vSLAM and generate multi-threaded C/C++ code.
交互式学习
图像处理入门之旅
该交互式免费教程时长不超过 2 小时，介绍了在 MATLAB® 中进行图像处理的一些实用技巧。
计算机视觉入门之旅
了解如何使用 Computer Vision Toolbox™ 进行目标检测和跟踪。
视频
什么是计算机视觉
了解计算机视觉如何扩展图像处理以实现各种应用领域，例如目标检测、跟踪和识别。
什么是 Medical Imaging Toolbox
探索 Medical Imaging Toolbox™ 的功能，包括可视化、注册、分割和标注。
