Single-Chip Multispectral Object Detection via Phase Detection Pixels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing imaging systems with object detection capabilities face challenges such as reduced frame rate, increased power consumption, and complexity due to the addition of components like frame memory, particularly in applications requiring natural lighting and LED lighting for face recognition.
Innovation Solution
The implementation of phase detection pixel blocks with asymmetric angular responses and band pass filters in image sensors, allowing for object detection without the need for frame memory by using a single image frame and optimizing lens systems for improved focus and power efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If frame memory is added to store the first frame image for subtraction operations, then object detection capability is improved, but device complexity increases
Solution Approach 1:
The patent extracts the frame memory component from the system by implementing a single-frame detection approach. The object detection is achieved by processing a single image frame captured under LED illumination, eliminating the need to store and subtract a second frame, thereby removing the frame memory requirement while maintaining detection capability
Solution Approach 2:
The image sensor is designed to perform multiple functions: capturing visible light images for display and simultaneously detecting objects through LED-illuminated frame analysis. This multi-functionality allows the same hardware to serve both imaging and object detection purposes without requiring separate dedicated components
2Difficulty of detecting and measuring
If frame memory is added to store the first frame image, then object detection capability is improved, but power consumption increases
Solution Approach 1:
The frame memory component is removed from the system architecture, eliminating the power consumption associated with maintaining and managing stored frame data. The single-frame approach processes only one image capture per detection cycle, reducing overall power usage compared to multi-frame storage and subtraction operations
3Difficulty of detecting and measuring
If frame memory is added for frame subtraction operations, then object detection capability is improved, but frame rate decreases
Solution Approach 1:
By removing the frame memory and multi-frame subtraction process, the system reduces the time required for each detection cycle. The single-frame approach captures and processes only one image per object detection event, eliminating the time penalty associated with capturing, storing, and subtracting multiple frames, thereby improving frame rate and processing speed
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This solution enhances object detection capabilities while reducing power consumption and complexity, improving frame rate and cost-effectiveness by eliminating the need for frame memory and allowing continuous low-intensity LED operation.
Implementation Method 1
Each pixel receives incident photons (light) and converts the photons into electrical signals
Data Source
AI summary
An imaging system may include an array of pixel blocks that each have multiple pixels over which a single microlens is formed. A lens system may be formed over the multiple pixels. The lens system may have an aperture and a filter system, which includes multiple portions each corresponding to a pixel of the multiple pixels, formed at the aperture. The multiple portions may have multiple band pass filters of different wavelengths. During imaging operations, a light emitting diode of a given wavelength may be used to brighten an object for detection. A band pass filter may correspond to the given wavelength of the light emitting diode. Subtraction circuitry may subtract signals obtained using a band pass filter from signals obtained using the band pass filter corresponding to the given wavelength to generate object detection data.