Digital Zoom via Sensor Cropping for Live Video
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Head-mounted display (HMD) systems face challenges in providing clear zoomed images of specific details in live video streams due to resolution limitations and bandwidth constraints, making it difficult for remote experts to read small text or barcodes, especially in poor lighting conditions.
Innovation Solution
The method involves capturing a second electronic image at the full resolution of a user-selected area of the sensor, rather than zooming into the original image, and communicating it at a lower resolution, thereby maintaining higher detail and improving lighting with variable pixel binning and correcting lens distortions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If digital zoom is applied to the streamed video, then the remote expert can view a larger area, but the image quality becomes blurry due to resolution limitations
Solution Approach 1:
The patent segments the sensor's field of view to capture a cropped image of only the user-selected area at full resolution. Instead of digitally zooming into a lower-resolution streamed frame, the system directs the sensor to capture a new image focused solely on the region of interest, thereby maintaining full resolution and image quality while providing the desired zoomed view to the remote expert.
Solution Approach 2:
The patent transitions from two-dimensional digital zooming (enlarging pixels within the same resolution plane) to a different dimensional approach by capturing a new image at full resolution focused on the selected area. This dimensional shift from post-processing zoom to targeted full-resolution capture eliminates blurriness while maintaining viewing area expansion.
2Measurement precision
If the HMD user moves closer to the machine to see details, then the image quality improves, but safety hazards increase due to moving parts or high voltage
Solution Approach 1:
The patent introduces an intermediary system consisting of the wearable device's sensor and the remote mentor engine that captures and transmits high-resolution images of the area of interest. This intermediary allows the remote expert to view detailed images without the HMD user physically moving closer to hazardous areas, thereby maintaining image quality while eliminating safety risks from proximity to moving parts or high voltage.
3Measurement precision
If camera mode is activated to capture full resolution image, then image quality improves, but the video stream must be interrupted causing workflow disruption
Solution Approach 1:
The patent implements a dynamic system where the sensor can switch between streaming mode and cropped full-resolution capture mode without complete interruption. The remote mentor engine dynamically processes user selections and triggers targeted full-resolution captures only when needed, allowing the video stream to continue while seamlessly integrating high-quality detailed images when the remote expert requests them.
4Measurement precision
If full resolution video is streamed continuously, then image quality is maintained, but bandwidth requirements exceed current limitations
Solution Approach 1:
The patent applies local quality enhancement by capturing full-resolution images only for the specific area of interest selected by the remote expert, rather than streaming the entire scene at full resolution. The sensor directs its full resolution capability locally to the cropped region, while the rest of the video stream maintains lower resolution, thereby achieving high image quality where needed while staying within bandwidth limitations.
Data Source
AI summary
Systems and methods facilitate digital zoom of live video based on remote user instructions. A user of a remote device may desire to zoom in on a particular aspect of a copy of a full image captured by a sensor at a first resolution and communicated to the remote device at a second resolution less than the first resolution. A selection that corresponds to a user-selected area of the copy and corresponding to a portion of the sensor is received by the sensor. Based on the selection, an instruction is generated to capture a second electronic image utilizing the portion of the sensor with the first resolution. Thus, rather than zooming into the first electronic image at the second resolution, the second electronic image corresponding to the portion of the sensor is captured at the first resolution and then converted to the second resolution, resulting in a higher resolution zoom.