In-Storage Image Processing Using Luminance-Chrominance Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional image processing methods using RGB data for machine learning algorithms require more computations, leading to increased time and power consumption, which is inefficient for applications like autonomous vehicles and smart stores that need quick object recognition.
Innovation Solution
Implementing a two-stage image processing method within a storage device using luminance (luma) data first, followed by chrominance data only when necessary, with machine learning models trained on luminance and chrominance data, respectively, to reduce computational load and power usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional image processing methods using RGB data are used for machine learning algorithms, then comprehensive color information is obtained, but computation time and power consumption increase significantly
Solution Approach 1:
The patent segments the image data into luminance (Y) and chrominance (CbCr) components, processing them through separate machine learning models. The luminance model processes grayscale information first for rapid object detection, while the chrominance model handles color information only when needed for differentiation, thereby reducing overall computation time while preserving necessary color information.
Solution Approach 2:
The system dynamically selects which model to execute based on the image characteristics and processing requirements. For images where color is not critical for identification, only the lighter luminance model runs. When color differentiation is needed, the system activates the chrominance model, optimizing the balance between information retention and processing speed.
2Reliability
If traditional image processing methods using RGB data are used for machine learning algorithms, then accurate object recognition is achieved, but power consumption increases
Solution Approach 1:
The patent divides the processing workload into two separate models: one for luminance and one for chrominance. The luminance model, which requires less computational power, handles the bulk of object detection. The chrominance model is invoked only when color information is necessary for accurate recognition, thereby reducing overall power consumption while maintaining recognition accuracy.
Solution Approach 2:
The system changes the data representation parameters by converting RGB images to YCbCr color space and separately processing the luminance and chrominance components. This parameter transformation allows the system to work with reduced data complexity for the primary detection task, lowering power consumption while preserving the ability to access full color information when needed.
3Loss of information
If RGB data with three channels is used for image processing, then complete color information is available, but the number of computations increases
Solution Approach 1:
The patent segments the three-channel RGB data into luminance and chrominance components, processing them through separate specialized models. The luminance model handles the primary detection task with reduced computational requirements, while the chrominance model supplements with color information only when necessary, thereby improving processing efficiency without losing complete color information.
Solution Approach 2:
The system applies partial action by processing only the luminance component for basic object detection, and adds chrominance processing only when the task requires color differentiation. This partial processing approach significantly reduces the number of computations compared to processing all three RGB channels simultaneously, while still providing complete color information when needed.
Data Source
AI summary
A programmable data storage device configured to process images via an embedded processor is disclosed. The processor identifies luminance and chrominance data of a received image, and retrieves a first machine learning model stored in the storage device. The first model is applied for making a first prediction about the image based on luminance data, and a first determination is made in regards to a criterion. In response to making the first determination, a first label associated with the first prediction is returned. A second determination is also made in regards to the criterion. In response to making the second determination, a second machine learning model stored in the storage device is retrieved. The second machine learning model is applied for making a second prediction about the image based on the color data associated with the image, and a second label associated with the second prediction is determined.