Jason Meltzer

University of California System, University of California, Los Angeles

Papers

Total Citations

H-Index

About

Jason Meltzer’s research sits at the intersection of computer vision and robotics, with a focus on enabling machines to perceive and navigate their environments. His key contributions lie in developing robust feature descriptors and applying them to the critical problem of Simultaneous Localization and Mapping (SLAM). In his most cited work, "Multiple View Feature Descriptors from Image Sequences via Kernel Principal Component Analysis" (2004, 40 citations), Meltzer pioneered a method to create richer, more invariant feature descriptors by leveraging multiple views of a scene. This work directly addressed the challenge of wide baseline matching under varying illumination and viewpoint. He then applied these descriptors to vision-based SLAM in his 2005 paper (16 citations), proposing a system that could build and maintain a map while tracking a robot’s position within it. His earlier work on "Vision Based Navigation" (2003, 5 citations) laid the groundwork by introducing a method for localizing a robot against a pre-built 3D map of image patches. Meltzer’s contributions are foundational for autonomous systems that must operate reliably in complex, real-world environments.