AI Sensor

AI Sensor algorithm solidification ability

TetrasMobile AI Sensor Hardware Capabilities

By leveraging the arrangement and physical characteristics of sensor pixels, coupled with the aid of AI algorithms, we address the SOOC image quality pain points of the front-end sensor, such as denoising, clarity improvement, bad pixel removal, etc., thus giving the sensor full play across multiple dimensions such as signal-to-noise ratio, color, HDR, and omni-directional focusing.

Return to List

Technical advantages

High Efficiency and High Definition

We apply our highly efficient, lightweight AI technology to enhance image quality in small sensors. Our approach overcomes challenges such as limited sensor area, power consumption standards, and calorific value controls. Our techniques include AI-RMSC, bad pixel removal, moire pattern removal, HDR fusion, color fusion, RAW domain deblurring, and depth map enhancement.

Controllable and Adjustable

Our AI Sensor Algorithm is controllable and adjustable. It combines physical property-based imaging models with machine learning, unlike the black-box nature of ordinary AI algorithms. This provides intuitive and robust parameters to adjust and tune image quality, while adapting to a wide range of sensors and ISP chip platforms. The algorithm ensures excellent image quality and full compliance with traditional sensor standards related to parameter tunability.

Hardware-Friendly

Our AI Sensor Algorithm RTL design and implementation capability addresses challenges in sensor RAW data processing, including storage, power consumption, and RAM reading and writing. It meets the performance requirements of image sensor hardware IP design while ensuring image quality. Tetras.AI has successfully applied sensor hardware IP design in flagship mobile phone models, including RGBW and 224OCL.