3D sensors can be mainly divided into the following categories: binocular camera, structured light, ToF camera, photometric stereo vision and stereo vision.
Binocular camera
Based on the principle of parallax, binocular cameras use two or more cameras to shoot the same scene from different positions, and calculate the position deviation between corresponding points of the image to obtain the three-dimensional geometric information of the object. Binocular cameras are divided into passive binocular and active binocular. Passive binocular uses visible light and does not require additional light source but cannot be used at night; active binocular uses infrared laser for fill light, which is suitable for scenes with low light. The advantages of binocular cameras include low hardware requirements and applicability to indoor and outdoor scenes, but the disadvantages are sensitivity to ambient light, high computational complexity, and not suitable for monotonous scenes lacking texture.
Structured light
Structured light technology uses invisible infrared laser of a specific wavelength as a light source, projects a coded pattern on the object, and calculates the distortion of the returned coded pattern to obtain the position and depth information of the object. Structured light cameras can be divided into stripe structured light, coded structured light and speckle structured light. Its advantages include mature solutions, convenient miniaturization, low resource consumption, high accuracy and high resolution, but its disadvantages are that it is easily disturbed by ambient light, poor outdoor experience, and the accuracy will deteriorate when the detection distance increases.
ToF camera
ToF camera calculates the distance of the target by measuring the flight time of infrared light in the air. The advantages of ToF technology are simple structure, easy to use, independent of ambient light, and suitable for application scenarios with a wide measurement range. Its disadvantages are that it can only provide ideal acquisition results under specific environmental conditions, limited measurement range, and may not be reliably measured under certain circumstances (such as highly reflective surfaces or dark surfaces).
Photometric stereo vision and stereo vision
Photometric stereo vision: Based on the known reconstruction function and the ideal Lambertian body reflection model, the image of the object under different light sources is captured by changing the direction of the light source to calculate the surface normal and three-dimensional shape of the object. Its equipment is simple but has strict requirements on the environment, and is suitable for smooth diffuse reflection surfaces.
Stereo vision: Simulating the principle of human vision, using two or more cameras to capture multiple images of the same scene from different angles, and measuring distance through stereo disparity. Its advantage is the simple hardware structure, but its disadvantage is that stereo matching is difficult, it is easily affected by occlusion or shadows, and it is difficult to accurately reconstruct objects without obvious surface features.
