Method and Device for Joining a Plurality of Individual Digital Images into a Total Image

Simple SummaryContent extracted from patent full text and abstract with AI.

This invention provides a method and device for combining multiple individual digital images into one comprehensive total image, using a feature-based algorithm. The process involves detecting features in the first image, tracking them in a subsequent image, and then calculating a transformation matrix from the matched features for precise image alignment. The system is robust enough to handle image distortion, lighting artifacts, and poor image quality, and works efficiently enough for real-time applications, such as in endoscopy, by leveraging algorithms like KLT (Kanade-Lucas-Tomasi) and incorporating outlier detection (RANSAC).

Use CasesContent extracted from patent full text and abstract with AI.

• Medical endoscopy: creating a wide-view map from sequential endoscopic images for enhanced diagnostics and surgical navigation.
• Industrial inspection: assembling panoramic images from borescopes for visual inspection of hard-to-access equipment or spaces.
• Surveillance or security: stitching images from security cameras to build composite overviews of large areas.
• Robotics and autonomous vehicles: generating environmental mosaics for navigation or mapping.
• Aerial or drone imaging: creating large-scale maps or orthophotos from overlapping aerial pictures.
• Augmented reality applications: real-time stitching of various camera feeds for immersive environments.
• Documentation and monitoring: recording and visualizing complete surfaces or interiors for maintenance or forensic purposes.

BenefitsContent extracted from patent full text and abstract with AI.

• Expands the effective field of view of imaging systems, notably endoscopes and borescopes, facilitating better orientation and analysis.
• Works in real-time, supporting dynamic or live applications, thanks to efficient feature-based algorithms and robust matrix calculation.
• Handles poor quality and distorted images robustly, making it suitable for challenging visual environments.
• Reduces manual intervention in combining images by automating feature detection, matching, and stitching.
• Minimizes the impact of artifacts such as glare or reflections by dynamically masking these in the computation process.
• Scales to multiple image sources and various image overlap configurations without loss of accuracy.
• Enables accurate documentation and measurement within the stitched total image, important for both medical and industrial records.

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

In a device (11) and a corresponding method for joining a plurality of individual digital images (Ia, Ib) into a total image (G), a plurality of features in a first individual image (Ia) are determined by means of a selection unit (15) using a feature-based algorithm and subsequently tracked in a second individual image (Ib) by means of a tracking unit (16). In a transformation unit (17), a transformation matrix is calculated from the determined feature correspondences, which is used to join the individual images (Ia, Ib) in an output unit (18) into a total image (G). The individual images (Ia, Ib) can be joined in real time and with high accuracy using the feature-based algorithm, combined with a robust algorithm for calculating the transformation matrix.