Human Image Recognition through Dynamic Mapping Generation

This paper proposed an image preparing technology to remove completely the noise effects of unexpected conditions and actually extract primary feature of human pose in indoor environment. Color image conversation with YCbCr, enhanced image intelligibility with a median filter, one OPEN operation and one labeling algorithm, are applied to detect the object‘s boundary. A dynamic image segmentation matching algorithm generated human feature vectors, which are used to search the most similar pattern for approximating the correct human pose. Several experiments show that the proposed dynamic image segmentation matching algorithm actually detected the real characteristics of human pose and makes the great support to re-cover the various image feature problem in complex environment. Future work for an intelligent human pose is used to virtually control home equipment in everywhere.