In-Sensor Face Detection Using Vertical Stripe Scanning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern cameras require large and area-consuming framestores to capture and analyze scenes, leading to substantial processing time and power consumption for identifying regions of interest, such as faces, due to the need to store and process entire frames.
Innovation Solution
Implementing an image sensor with a detector that scans the scene in successive time intervals using vertical stripes, storing only a portion of the scene in a smaller image buffer, and using a rolling buffer for face detection, allowing for efficient detection and processing of faces without needing to store the entire frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a large framestore is used to store the entire scene, then complete scene data is available for analysis, but the device area and memory size increase substantially
Solution Approach 1:
The patent extracts only the essential information needed for object detection (face presence indicators) from the complete scene data, storing merely the presence/absence flags in a compact buffer rather than the entire frame data. This extraction principle reduces the stored data from full-frame resolution to minimal detection markers.
Solution Approach 2:
Instead of storing the actual scene data, the system creates simplified copies in the form of presence/absence indicator buffers that represent the essential detection information without duplicating the full scene data, thereby reducing memory requirements while preserving detection capability.
2Measurement precision
If the entire frame is stored and analyzed by a high-power processor, then accurate object identification is achieved, but processing time and power consumption increase
Solution Approach 1:
The system performs preliminary scanning of the scene during the imaging process itself, using the imaging sensor to detect face presence indicators before full frame processing is required. This preliminary detection action occurs concurrently with scene capture, eliminating the need for subsequent full-frame analysis.
Solution Approach 2:
The patent implements continuous face detection scanning throughout the scene capture process, maintaining detection operations continuously rather than performing batch processing after frame capture. This continuous action allows detection to proceed in parallel with imaging, maximizing utilization of the imaging process for dual purposes.
3Measurement precision
If the entire frame is stored and analyzed by a high-power processor, then accurate object identification is achieved, but power consumption increases
Solution Approach 1:
The system extracts only the minimal necessary data (face presence indicators) from the complete scene, avoiding the energy-intensive operation of processing full-frame data. This extraction reduces computational load and associated power consumption while maintaining detection accuracy.
Solution Approach 2:
The patent creates simplified indicator copies instead of processing full-frame data, using low-power buffer operations to store and retrieve presence/absence flags rather than performing high-power processor operations on complete image data, thereby significantly reducing energy consumption.
4Area of stationary object
If a smaller image buffer is used to store only a portion of the scene, then device area is reduced, but complete scene coverage requires multiple scans
Solution Approach 1:
The system employs periodic scanning of the scene in successive time intervals, systematically working through different portions of the scene across multiple scan cycles. This periodic scanning approach ensures complete scene coverage is achieved over time while using minimal buffer memory at any given moment.
Solution Approach 2:
The patent introduces the time dimension to the scanning process, transforming a spatial storage problem into a temporal processing solution. By scanning different scene portions across successive time intervals rather than loading the entire scene simultaneously, the system achieves complete coverage with minimal buffer requirements.
Data Source
AI summary
Systems and methods are provided for detecting an object of object class, such as faces, in an image sensor. In some embodiments, the image sensor can provide a scan sequence that scans a scene over multiple time intervals. The image sensor can scan, in succession, portions of a scene, where each of the portions covers a different amount or location of the scene. This way, the scanned portions can be saved in an image buffer that is sized significantly smaller than an entire frame. In some embodiments, when the image sensor detects the presence of an object of the object class, the image sensor can store positional information (e.g., location and size of the object) in a region of interest buffer. The image sensor can output the positional information to aid an electronic device, such as a camera, perform various functions, such as automatic exposure and color balancing.