Depth Resolution Enhancement Technique for CMOS Time-of-Flight 3-D Image Sensors

Abstract

Introducing Time-of-Flight 3-D image sensors to actual engineering applications, such as pattern recognition, is constrained not only by their limited depth and lateral resolution, but also by how similar the precision of depth measurement throughout the whole pixel-matrix is. In real operating environment, an observed 3-D-scene hardly exhibits a homogeneous reflectance factor. Moreover, the light-beam (laser source) presents a nonuniform optical power distribution in space. Thus, the amount of the incident light on the sensor surface varies drastically from one pixel to another, and so does the signal-to-noise ratio. To address this problem, this paper investigates the impact of both scene and light-source non-ideal characteristics on the sensor performance. An adaptive on-pixel analog signal processing technique is also presented and applied to the design of a 32 × 32 complementary metal oxide semiconductor (CMOS) range camera, featuring an interesting cost-efficient solution.