Object Location and Track in Image Sequences by Means of Neural Networks

Object location and track in image sequences is an important task in computer vision, which has many applications. Major challenges of object track have been, and continued to be, improvement of its accuracy and real-time performance. In this paper, a novel BPneural-network-based object location approach is proposed, in which a threshold for the object-matching quality is used for determining whether the object is present in a given frame. To simplify the network structure, a directional wavelet transform (DWT) is used for extracting image features, which can reduce the size of the input patterns. In order to further improve the computation speed of the method, the information on the position of the target object in the previous frames is used for predicting the position of the target object in the current †* This work was supported by Hori Information Science Promotion Foundation. *† Corresponding Author. Email: zhshi@ieee.org. International Journal of Computational Science 1992-6669 (Print) 1992-6677 (Online) www.gip.hk/ijcs © 2008 Global Information Publisher (H.K) Co., Ltd. 2008, Vol. 2, No. 2, 274-285. International Journal of Computational Science GLOBAL INFORMATION PUBLISHER 274 frame. Experiments indicate that the proposed method is more accurate in target detection and more computationally efficient than conventional methods.

[1]  Narendra Ahuja,et al.  Extraction of 2D Motion Trajectories and Its Application to Hand Gesture Recognition , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[2]  Bohyung Han,et al.  On-line density-based appearance modeling for object tracking , 2005, Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1.

[3]  Benjamin W. Wah,et al.  Global Optimization for Neural Network Training , 1996, Computer.

[4]  Linhua Zhang,et al.  Hopfield Neural Network Image Matching Based on Hausdorff Distance and Chaos Optimizing , 2005, ISNN.

[5]  Roberto Battiti,et al.  Accelerated Backpropagation Learning: Two Optimization Methods , 1989, Complex Syst..

[6]  Wen-Jing Li,et al.  Hopfield neural networks for affine invariant matching , 2001, IEEE Trans. Neural Networks.

[7]  Franco Bartolini,et al.  Motion estimation and tracking for urban traffic monitoring , 1996, Proceedings of 3rd IEEE International Conference on Image Processing.

[8]  William Finnoff,et al.  Diffusion Approximations for the Constant Learning Rate Backpropagation Algorithm and Resistance to Local Minima , 1992, Neural Computation.

[9]  Geoffrey E. Hinton,et al.  Learning representations by back-propagation errors, nature , 1986 .

[10]  Nasser M. Nasrabadi,et al.  Object recognition by a Hopfield neural network , 1990, [1990] Proceedings Third International Conference on Computer Vision.

[11]  Patrick Bouthemy,et al.  Computation and analysis of image motion: A synopsis of current problems and methods , 1996, International Journal of Computer Vision.

[12]  Jung-Hua Wang,et al.  On Disparity Matching in Stereo Vision via a Neural Network Framework , 1999 .

[13]  Jake K. Aggarwal,et al.  On the computation of motion from sequences of images-A review , 1988, Proc. IEEE.

[14]  Robert A. Jacobs,et al.  Increased rates of convergence through learning rate adaptation , 1987, Neural Networks.

[15]  David Beymer,et al.  A real-time computer vision system for vehicle tracking and traffic surveillance , 1998 .

[16]  Guoyou Wang,et al.  Relaxation matching by the Hopfield neural network , 1996, Electronic Imaging.

[17]  Narendra Ahuja,et al.  Detecting Faces in Images: A Survey , 2002, IEEE Trans. Pattern Anal. Mach. Intell..