Face Detection Using Pseudo-Distance Scaling for Out-of-Focus Images

The patent performs preliminary action by calculating the pseudo-distance between the image pickup module and the object before face detection. This preliminary distance estimation allows the system to determine an appropriate scaling value in advance, which is then used to scale the image before face detection. This resolves the contradiction by enabling the system to handle out-of-focus images (improving adaptability) while maintaining detection accuracy through pre-calculated scaling parameters.

The patent introduces a pseudo-distance calculation as an intermediary step between image acquisition and face detection. This intermediary parameter serves as a mediator that bridges the gap between the out-of-focus image state and the face detection process. By using the pseudo-distance to determine scaling values, the system can accurately detect faces in out-of-focus images without requiring the image to be in focus, thus resolving the accuracy-adaptability contradiction.

The patent applies parameter changes by using the pseudo-distance to dynamically determine an optimal scaling value rather than testing multiple fixed scaling values. The pseudo-distance parameter changes the scaling value selection process from a brute-force approach to a calculated approach. This reduces processing time while maintaining accuracy because the scaling value is derived from the actual distance information rather than being guessed or tested exhaustively.

The patent replaces the mechanical system of testing multiple scaling values (which would be like trying different physical configurations) with a computational substitution using pseudo-distance calculations. Instead of mechanically testing multiple scaling options, the system calculates the appropriate scaling value based on the pseudo-distance parameter. This substitution dramatically reduces processing time while achieving the same or better detection accuracy.