Friday 21 March 2025
Scientists have made a significant breakthrough in the field of computer vision, developing a new method that allows them to capture high-quality images of objects without using a traditional studio setup. The innovative approach uses a mobile phone as a light source and camera, enabling photometric stereo reconstruction – a technique used to estimate the surface normal map of an object.
Photometric stereo is typically performed in a controlled environment with specialized equipment, such as a large LCD monitor and a polarization camera. However, this new method, known as Differentiable Mobile Display Photometric Stereo (DMDPS), can be done using a standard mobile phone, making it more accessible and practical for everyday applications.
The researchers achieved this by developing a custom app that allows them to display patterns on the mobile phone’s screen and capture images of an object using the front camera. The patterns are designed to illuminate the object from different angles, allowing the scientists to reconstruct its surface normal map.
One of the key challenges in photometric stereo is separating diffuse and specular reflections on the object’s surface. In traditional setups, this is achieved using a polarization camera and LCD monitor. However, DMDPS uses a mobile phone as a light source, which limits the ability to separate these reflections. Despite this limitation, the researchers were able to achieve high-quality surface normal reconstructions by using HDR imaging and undistortion techniques.
The team tested their method on various objects, including 3D-printed models and natural objects like fallen leaves. The results show that DMDPS can accurately capture the surface normal map of an object, even in complex environments with varying lighting conditions.
This breakthrough has significant implications for computer vision and robotics. With DMDPS, scientists can now use mobile phones to quickly and easily capture high-quality images of objects, enabling applications such as augmented reality, robotics, and quality control.
The researchers believe that their method could be used in a variety of fields, including manufacturing, healthcare, and entertainment. For example, in manufacturing, DMDPS could be used to inspect the surface quality of products or monitor the wear and tear of machinery. In healthcare, it could be used to analyze the texture and structure of biological tissues.
The development of DMDPS demonstrates the potential for mobile devices to play a more significant role in computer vision research. As mobile phones become increasingly powerful and versatile, scientists are exploring new ways to harness their capabilities to advance our understanding of the world around us.
Cite this article: “Mobile Phone-Based Photometric Stereo Reconstruction”, The Science Archive, 2025.
Computer Vision, Photometric Stereo, Mobile Phone, Surface Normal Map, Light Source, Camera, Hdr Imaging, Undistortion Techniques, Robotics, Augmented Reality.
