ACCURATE HEIGHT MEASUREMENTS IN PERSPECTIVE EMPLOYING THREE VANISHING POINTS FROM A SINGLE IMAGE

Background: This study explores the need to obtain accurate measurements from single images, applicable in areas such as architectural reconstruction and forensic investigations. The technique uses vanishing point geometry to deduce real measurements from known references. Aim: The main objective is to demonstrate a method for measuring the height of objects and people in images using three vanishing points, allowing the analysis of real dimensions from an image. Methods: This article presents a method for accurately measuring height in images using vanishing points and cross-ratio geometry. By using three vanishing points corresponding to orthogonal axes, the technique allows the recovery of real-world dimensions from a single image. Vanishing points, which represent the intersection of parallel lines in space when projected onto a 2D plane, allow for precise spatial mapping and dimensional analysis. In this study, known references were used for calibration. Results and Discussion: The results demonstrated that the methodology is reliable, producing height estimates between 1,86 and 1,92 meters for a subject who is 1,91 meters tall, with deviations within acceptable margins of error. These findings highlight the robustness of the method, especially in applications where direct measurements are not feasible, such as forensic investigations or architectural reconstructions. The accuracy of the methodology was validated through repeated tests, where 10 iterations showed a concentration of results within a narrow range, further reinforcing its reliability. Conclusions: This technique offers a practical solution for extracting precise measurements from images, with potential applications in architecture, forensic science, and virtual/augmented reality. The results suggest that, with continued refinement, the accuracy of this approach could be improved, making it a valuable tool in several fields that require spatial analysis.