Z-Order Image Buffer for Face Detection in Sensors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern cameras require large and area-consuming framestores to store entire frames for object detection, leading to significant processing time and power consumption.
Innovation Solution
Implementing an image buffer that stores data in raster order and reads it out in Z-order, allowing for efficient detection of objects like faces using a detector configured with feature response calculators and accumulator/decision modules, which minimizes the need for a framestore and reduces processing time by enabling concurrent processing of image data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large framestore is used to store entire frames for object detection, then object detection capability is improved, but device area and memory size increase
Solution Approach 1:
The patent extracts only the necessary portion of image data (e.g., 2x2 pixel blocks) for object detection rather than storing entire frames. This selective extraction allows face detection algorithms to operate on minimal data, eliminating the need for large framestores while maintaining detection accuracy.
Solution Approach 2:
The patent segments the image into small 2x2 pixel blocks and processes them independently for face detection. This segmentation enables the system to detect faces using only local pixel relationships rather than requiring global frame context, significantly reducing memory requirements.
2Measurement precision
If entire frames are stored and repeatedly fetched for analysis, then object detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary processing by immediately converting incoming pixel data into 2x2 blocks and computing face detection metrics on-the-fly. This preliminary action eliminates the need for subsequent repeated fetching and re-analysis of entire frames, reducing processing time while maintaining detection accuracy.
Solution Approach 2:
The patent enables continuous face detection by processing pixel blocks as they arrive from the sensor without interruption. The detector continuously analyzes 2x2 blocks in real-time, maintaining uninterrupted object detection capability without the time losses associated with frame-by-frame processing.
3Measurement precision
If a dedicated framestore is implemented, then object detection function is improved, but power consumption increases
Solution Approach 1:
The patent extracts only essential 2x2 pixel block data for face detection rather than maintaining large framestore memory. This extraction approach minimizes the active memory components, directly reducing the power consumption associated with framestore operations while preserving object detection functionality.
Solution Approach 2:
The patent uses partial action by processing only the minimum necessary data (2x2 blocks) for face detection rather than analyzing entire frames. This partial processing approach significantly reduces computational load and associated power consumption while maintaining sufficient detection accuracy for the application.
4Measurement precision
If high-power processors are used for repeated frame analysis, then object detection capability is improved, but device complexity and area increase
Solution Approach 1:
The patent applies partial action by implementing a simplified detector that processes only 2x2 pixel blocks rather than requiring high-power processors to analyze entire frames. This approach uses minimal computational resources to achieve face detection, significantly reducing device complexity while maintaining detection capability.
Solution Approach 2:
The patent uses a simplified computational model that copies the essential face detection functionality from complex frame-based algorithms to a lightweight block-based processor. This copying approach enables object detection using simple logic operations on 2x2 blocks, eliminating the need for high-power processing hardware.
Data Source
AI summary
Systems and methods are provided for detecting objects of an object class, such as faces, in an image sensor. In some embodiments, the image sensor can include a detector with an image buffer. The image buffer can store image data in raster order. The detector can read the data out in Z order to perform object detection. The image data can then compute feature responses using the Z-ordered image data and determine whether any objects of the object class are present based on the feature responses. In some embodiments, the detector can downscale the image data while the object detection is performed and use the downscaled image data to continue the detection process. In some embodiments, the image data can perform detection even if the image is rotated.