Image Processing Device Object Detection Rate Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems for surveillance cameras face challenges in efficiently detecting and encoding object data from image frames, particularly in managing bit rates and quantization parameters to balance image quality and object mobility, while maintaining real-time processing and accurate object detection.
Innovation Solution
An image processing device that detects object data from first image frames at lower intervals than the frame rate, estimates object data for missing frames based on previous data, and encodes frames at the sensor's frame rate, using different quantization parameters and bit rates for regions of interest and non-interest, with adjustments based on object mobility and image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If object data is detected from every image frame at the sensor's frame rate, then object detection accuracy is improved, but processing complexity and computational load increase significantly
Solution Approach 1:
The patent applies preliminary action by detecting object data only from selected first image frames at predetermined intervals rather than every frame, and then estimating object data for subsequent second image frames based on the detected data and object movement patterns. This reduces the frequency of actual detection operations while maintaining continuous object tracking, thereby lowering processing complexity while preserving detection accuracy.
Solution Approach 2:
The patent uses copying by estimating object data for second image frames based on object data from first image frames. Instead of performing full object detection on every frame, the system creates estimated copies of object data for intermediate frames by extrapolating from detected data and movement vectors, significantly reducing computational load while maintaining detection effectiveness.
2Ease of manufacture
If uniform quantization parameters are applied to all regions of image frames, then encoding simplicity is maintained, but image quality in regions of interest deteriorates
Solution Approach 1:
The patent applies local quality by setting different quantization parameter values for different regions of the image frame. Specifically, regions of interest (where objects are detected) receive lower quantization parameter values to maintain higher image quality, while non-interest regions receive higher quantization parameter values for greater compression. This regional differentiation resolves the contradiction between encoding simplicity and image quality by applying quality optimization only where necessary.
3Manufacturing precision
If high bit rate is used for encoding all image frames, then image quality is improved, but bandwidth consumption and storage requirements increase
Solution Approach 1:
The patent applies local quality in the bit rate allocation by determining different bit rate values for regions of interest and non-interest regions. Regions containing detected objects are allocated higher bit rates to maintain quality, while empty or non-critical regions receive lower bit rates for compression. This resolves the contradiction by concentrating bandwidth resources only where quality is essential.
Solution Approach 2:
The patent uses parameter changes by dynamically adjusting quantization parameters and bit rate values based on the presence and characteristics of detected objects. When objects are detected in certain regions, the system changes the encoding parameters for those regions to preserve quality, while maintaining lower parameters in other regions. This dynamic parameter adjustment optimizes the trade-off between image quality and bandwidth consumption.
4Productivity
If object detection rate is reduced to lower processing load, then processing speed is improved, but object detection accuracy deteriorates
Solution Approach 1:
The patent applies preliminary action by performing actual object detection only on selected first image frames at reduced frequency (predetermined intervals), then using the results to estimate object presence and position in subsequent frames. This preliminary detection approach maintains processing speed by reducing detection frequency while preserving accuracy through intelligent estimation for intermediate frames.
Solution Approach 2:
The patent uses feedback by continuously updating object data estimates based on newly detected object data from subsequent first image frames. The system incorporates feedback from actual detection results to correct and refine estimated object positions and movements, ensuring that detection accuracy is maintained over time despite the reduced detection rate.
Data Source
AI summary
An image processing device including a processor configured to: detect object data of an object from first image frames, among image frames input from an image sensor, at predetermined intervals according to an object detection rate which is lower than a frame rate of the image sensor; estimate, based on the object data of the first image frames, object data from second image frames other than the first image frames among the image frames; and encode the image frames at an encoding frame rate that is equal to the frame rate of the image sensor, based on the object data.