Liveness Detection via Dynamic Facial Model Binding

The patent implements dynamic detection by continuously tracking the movement trajectory of a control element on the screen. Instead of static detection, the system captures multiple position points over time as the control moves, creating a dynamic detection process that records the complete movement path. This dynamic approach significantly increases detection accuracy by capturing temporal and spatial characteristics that static methods miss, while the automation of trajectory tracking keeps the operational complexity manageable.

The patent introduces a virtual control element (such as a slider or draggable object) as an intermediary between the user's facial movements and the detection system. The control's movement trajectory on the screen serves as a mediator that translates subtle facial muscle movements into measurable positional data. This intermediary amplifies the detection signals and provides a clear, quantifiable metric for determining liveness, thereby improving detection accuracy without requiring direct analysis of complex facial features.

The system employs dynamic trajectory tracking where the control element moves along a continuous path rather than executing discrete, pre-defined actions. The movement trajectory is captured as a sequence of position points over time, creating a unique temporal-spatial pattern for each genuine user. This dynamic approach generates highly variable detection data that is extremely difficult for machines to replicate through exhaustive attacks, thereby enhancing security while the automated tracking simplifies implementation.

The system implements continuous feedback by monitoring the real-time position of the control element throughout its movement trajectory. Each position point provides feedback information that contributes to the overall liveness determination. This continuous feedback mechanism allows the system to detect subtle deviations in movement patterns that indicate automated attacks, significantly improving security. The feedback loop is automatically processed through the detection algorithm, maintaining ease of implementation while enhancing robustness against malicious attempts.

The patent segments the continuous movement trajectory into discrete position points captured at preset time intervals. Each position point represents a segment of the overall movement path, and the complete trajectory is reconstructed by combining these segmented data points. This segmentation approach enables accurate capture of movement characteristics including speed, acceleration, and trajectory shape, while the modular nature of segmented data simplifies processing compared to analyzing continuous raw data streams.

The patent replaces complex mechanical analysis of facial movements with a simplified digital tracking system. Instead of using sophisticated sensors or mechanical apparatus to measure facial muscle movements, the system uses software-based tracking of a visual control element on the screen. This substitution of mechanical measurement systems with digital/image processing methods significantly reduces data processing complexity while maintaining or improving measurement precision, as the control's position can be easily extracted from screen coordinates.