Articulated Object Detection And Recognition Applications

CV aims at enabling artificial intelligent perception of input multi-dimensional image data. Image contents can be generated and understood by recognizing objects in the cluttered images; thus object detection/recognition becomes a fundamental researching field. An efficient object detection/recognition algorithm is a basic prerequisite in many subjects, including, but not limited to:Content-Based Images Retrieval(CBIR), Medical Image Processing, Analysis, and Visualization(MIPAV), digital video surveillance systems, interactive industrial robots. However, object recognition is still in its first preliminary stage of the rapid development, although no general yet robust framework of the algorithm theory in the articulated object is available.In this paper, we proposed a simple but efficient iterative object recognition algorithm which was presented based on the articulated object models. Improved models are used to search local matches in cluttered images backgrounds, to organize qualified matching assumptions. Background clutter was always a "annoying" factor to the recognition performance. SC was improved mainly to avoid the background clutter and gain better tolerance to shape deformation. This kind of finding matches using pre-defined object model is called top-down recognition process. It usually has a high recall rate but suffers from low recognition accuracy. To improve the detection/recognition accuracy, hypotheses are further identified using a discriminative classifier. Object foreground regions in the articulated images are finally obtained by combining the cues from bottom-up image segmentations information.The experiment results show that our selected shape feature can remove background clutter effectively, and achieve good detection and matching results.P.MTM is an important occasions for the articulated object detection. Meanwhile, it will be favorable for the online sales company and provide a more efficient sale channels. The proposed method can help to extract the human body feature points and physical parameters from the contour image.