Understanding images by recognizing its objects is still a challenging task. Tracking of moving human and recognition have been developed by researchers but not yet shows enough information needed for recognition. Initially, a tracking process of an object starts with detection and recognition of the object in a static pose and position, and then continues in movement in different poses. Available moving human recognition methods still has error in classification and need a huge amount of examples which may still be incomplete. Human face and body posture characteristics such as size of the eyes, nose, mouth, or fat or thin bodies, are important visual features in different poses for personal identification to increase accuracy of human recognition system, and it is still rare in researches. This paper attempts to describe visual features that best known for human, but hard to be recognized by machines. Curve fitting approaches to face and body posture features are also introduced to capture exact patterns of the features. Body postures are also preprocessed with a Kinect depth camera, and also compared to popular and recent methods of visual object recognition. Finally, we demonstrate our method can be useful for visual object classification. Probabilities of personal identification can be increased by using different poses and characteristics of smaller detail features through body postures and face areas. More detail features will richen comparison data samples for higher recognition accuracy.
Since the Viola and Jones' method on real-time face detection was proposed in 2001, numerous works for object detection, person recognition, and object tracking have been published by papers and journals. Each method has its strong points and drawbacks. That means that in a system which only employs a standalone method, we could only get either speed or accuracy. In this paper, we proposed a state-machine method to combine face recognition, face detection, and tracker to harness the tracker promptness while maintaining the ability to distinguish the person of interest with the other person and backgrounds, to overcome the limitations of the standalone method. Subsequently, the information gathered from this image processing side will be delivered to the hardware tracker. The image processing side becomes a visual sensor that provides feedback or measurement value i.e. center point coordinate value from the detected face. The 2 DOF hardware tracker camera platform being used implements Model Predictive Control to calculate required control action thus the platform is able to track the target object, keeping it at the center of the frame. MPC method is chosen because it produces an optimal control signal while considering the input signal saturation aspect. The MPC control signals deliver a good control pan and tilt system response with rise time < 1 second and overshoot <15%. It is also noticed that the FSM implemented in this paper is able to meet the goal with a considerable performance for indoor settings.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.