The invention of photography technology has changed the way people depict the world. Photography technology starting from silver salts has undergone several revolutionary developments from black and white to colour, from film to digital sensing. Computational photography represents the latest trend in the development of photography technology. Computational photography is a comprehensive technology that combines the computer and software methods with modern sensors, modern optics and other technologies to create new photographic equipment and applications. Traditional photography techniques, including film and digital photography, tend to focus on the design of components, especially lenses and image sensors, but computational photography can get rid of the limitation of traditional photography through the collaboration of software and hardware, and create novel image applications by the change of optics, sensors and image processing, so the computational photography technology is widely used in mobile devices, surveillance and automation equipment, machine vision systems, and has great potential in digital photography, visual arts, visual communication, social networks, digital entertainment, monitoring, multimedia and other fields.
Mobile terminal computational photography system refers to a system that creates novel image applications through imaging control, image computing and processing, and combining artificial intelligence and mobile internet technology to reconstruct scenes, light and objects relying on limited optical devices and computing units on the mobile terminal. With the large-scale popularization of smart phones, tablets and other mobile terminals, taking photos has become an important part of people's daily social contact and entertainment. However, due to the limited size of mobile terminals, the photography hardware based on them cannot be compared with the professional photographic equipment. Therefore, the computational photography system focusing on software and hardware cooperation and image processing algorithm has become an important way to improve the quality of photography and enrich the image function on the mobile terminal. The world's leading mobile terminal manufacturers take computational photography system as the main part of mobile phone for research and development. Through a series of techniques, including the scene perception, image semantic segmentation, image semantic understanding, real-time beauty camera, multiple cameras, multi-frame fusion, focus stacking, focusing after shooting, HDR, panorama stitching and so on, they bring cell phone users with extra sensual perception experience.
Due to the differences in hardware architectures and operating systems of different mobile terminals, the computational photography systems and photography applications running on them are often incompatible. In order to adapt different types of terminals, application developers have to develop multiple versions of their applications to ensure the consistency of user experience. This proposal attempts to establish a reference architecture of mobile terminal computational photography system. The Recommendation is aiming to improve the production efficiency, promote industrial cooperation and meet the demand of the technology development for the next step.