lngenic 君正

Gekko

Ingenic AISP technology

The visual perception industry has undergone long-term development, and traditional ISP image processing effects have been continuously optimized and upgraded. However, in the face of increasing scene complexity and higher image quality requirements, traditional ISP technology has approached the optimization limit. Stable output of high-quality images in complex scenes such as extremely low light, backlight, wide dynamic, and motion remains a challenge in the visual perception industry. In this context, breaking through the obstacles of image quality optimization by introducing AI technology into ISP has become an important research direction, and significant progress has been made, bringing a qualitative leap to stable and high-quality imaging in complex scenes.

In line with technological trends and industry trends, Ingenic has independently developed the AISP Intelligent Image Processing Engine (Gekko) based on its long-term accumulation in AI and ISP technologies. Compared to traditional ISPs, it has significantly improved the quality of real-time and high-definition video images, solved extremely low-light and full-color night vision imaging,improved contrast enhancement and ultra wide dynamic range imaging, all of these are difficult to solve in industries.

Full-color night vision imaging

Gekko introduces AI models trained by deep learning technology into key modules such as 3D noise reduction, low light enhancement, strong light suppression, and contrast enhancement in traditional ISP pipelines, achieving high-definition and full-color night vision imaging in various complex and variable extremely low light scenes, making the overall image bright and transparent, with rich and layered details, clean and gorgeous colors, solid and clear motion areas, and harmonious unity between static and dynamic areas of the image.

High dynamic range imaging

By introducing intelligent exposure control algorithms, Gekko can adaptively adjust multiple exposures for different scenes, capture different levels of light and shadow in different scenes, and obtain appropriate exposure values for both bright and dark parts of the image. Using intelligent fusion anti-ghost algorithms and AI-LTM algorithms to fuse and adjust long, medium, and short frames with different exposure times, the overall image is bright and transparent, with clearer layers, richer details, and more vibrant colors, and no obvious ghosting in the motion area.