Touch is one of the most intuitive ways for humans to interact with the world, and as we advance toward a ubiquitous computing environment where technology seamlessly integrates into daily life, natural interaction methods are essential. This paper introduces UbiTouch, a system leveraging thermal imaging to detect touch interactions on arbitrary surfaces. By employing a single thermal camera, UbiTouch differentiates between hovering and touch, detects multi-finger input, and completes trajectory tracking. Our approach emphasizes the use of lightweight, low-computation algorithms that maintain robust detection accuracy through innovative vision-based processing. UbiTouch aims to enable scalable, sustainable, and adaptable interaction systems for diverse applications, particularly with regards to on-human sensing.