Zebra Aurora Imaging Library Software

• Focus on solving application challenges using a toolkit that boasts over a 25-year track record of dependable performance.
• Confidently address applications with field-tested tools for analyzing, classifying, locating, measuring, reading, and verifying
• Base your analysis on a range of inputs including monochrome and color 2D images as well as 3D profiles, depth maps, and point clouds
• Utilize cutting-edge hardware optimizations leveraging SIMD, multi-core CPUs, and multi-CPU technologies
• Support a wide array of platforms from smart cameras to high-performance computing (HPC) clusters via an intuitive and unified API
• Capture live data through multiple interface supports such as analog, Camera Link®, CoaXPress®, DisplayPort™, GenTL, GigE Vision®, HDMI™, SDI, Linux® and USB3 Vision®
• Maintain system flexibility with support for 64-bit Windows® and Linux® systems along with Intel® and Arm® processors
• Leverage existing programming knowledge with language support for C, C++, C#, and CPython
• Experiment and produce program code using the Aurora Imaging Library CoPilot interactive setting
• Enhance productivity and reduce development costs through Vision Academy's training, available online and on-site

Industrial image processing tools Aurora Imaging Library, previously named Matrox Imaging Library, aggregates a comprehensive set of software tools vital for developing machine vision and image analysis applications. The Aurora Imaging Library spans the entire project lifecycle, from viability assessment to prototyping through to development and deployment. The SDK encompasses interactive software components and programming tools for image capture, processing, analysis, annotation, visualization, and archiving. These tools are engineered to maximize productivity and substantially reduce the time and effort required to market solutions. Initial release in 1993, the Aurora Imaging Library has continuously evolved to adapt to emerging industry demands. Its intuitive and consistent API has established a long-standing reputation for being user-friendly. A dedicated team continues to maintain and advance the Aurora Imaging Library adhering to industry-best practices, peer reviews, user engagement, and frequent updates.

Synopsis

Aurora Imaging Library is a comprehensive suite of software tools intended for the development of machine vision and image analysis applications. It combines interactive software solutions and programming functionalities for image acquisition, processing, analysis, annotation, display, and archiving to enhance productivity and minimize the time and effort needed to introduce solutions to the market. The Aurora Imaging Library is continually refined and improved, following industry-recognized best practices such as peer review, user feedback, and daily builds.

Deep Learning Model for Object Detection

Aurora Imaging Library includes Classification tools that employ machine learning for the automatic categorization of image content or previously extracted elements. These tools utilize deep learning technology—specifically convolutional neural networks (CNNs) and their variants—in three main approaches. The global approach classifies images or image regions into predefined categories, making it suitable for tasks where differentiation between similarly appearing objects is required. The segmentation approach creates maps showing the predefined class and score for all image pixels, aiding in the detection of flaws or features. The object detection approach identifies instances of predefined classes, optimal for finding, scaling, and counting objects or features. This classification is particularly effective for inspecting images of highly textured, naturally varying, and acceptably deformed goods in complex and varying environments. Users can train a deep neural network themselves or have Zebra do so using a sufficient number of representative images. Training methods such as transfer learning and fine-tuning are supported, starting from one of the provided predefined deep neural network architectures.

Deep Learning Inference on Both Intel and NVIDIA GPUs

Aurora Imaging Library supports deep learning inference on both Intel integrated GPUs and NVIDIA GPUs.

3D Surface Matching Tool

Aurora Imaging Library includes a tool for identifying surface models—including multiple occurrences—within a point cloud. Surface models can be created from point clouds sourced from 3D cameras/sensors or CAD files. Multiple controls are available to adjust search accuracy, robustness, and speed. Search outcomes include the number of occurrences discovered and for each one, the score, error, number of points, central coordinates, and estimated pose. This tool is ideal for locating and evaluating objects within a point cloud, subsequently permitting detailed analysis using 2D vision tools like Pattern Recognition and Character Recognition, as well as 3D tools such as Metrology.

3D Shape Finding Utility

Aurora Imaging Library offers a utility to identify specific shapes—such as cylinders, hemispheres, rectangular planes, and boxes—within a point cloud. Users can specify shapes numerically or through predefined shapes. The utility allows customization to fine-tune finding accuracy, robustness, and speed. Results detail the number of occurrences, score, error, number of points, and central coordinates, with further specifications for different shapes like radius or length.

3D Blob Analysis Utility

Aurora Imaging Library's 3D Blob Analysis utility is designed to segment and inspect objects within a point cloud. This utility detects blobs, measures numerous features, sorts and combines blobs for extensive analysis. It is especially useful for inspecting highly textured, naturally variable, and acceptably deformed items in complex scenes.

Feature Extraction and Analysis Utilities

Aurora Imaging Library provides essential tools for image analysis, including Blob Analysis and Edge Finder utilities. These tools identify and measure basic characteristics to determine object presence and location. Blob Analysis works on binary images to quickly detect and measure over 50 characteristics of blobs, employing run-length encoding for rapid processing. Edge Finder is excellent for scenes with fluctuating lighting, using gradient-based and Hessian-based methods to locate contours and ridges in images with high accuracy.

Image-Oriented Classification (Global Method)

Aurora Imaging Library includes Classification tools that facilitate the automatic categorization of image contents or pre-extracted features using machine learning. Employing deep learning technologies such as convolutional neural networks (CNNs), the library offers three key methods