Bio-inspired model for object recognition based on histogram of oriented gradients

Hierarchical Model and X (HMAX) is an outstanding bio-inspired model for object recognition. But there are still some limitations such as its poor invariance to rotation and processing speed. To lessen these limitations and improve the model's performance, we extend the HMAX model and propose a novel representation method by combining HMAX with Histogram of Oriented Gradients (HOG), denoted as HOG-HMAX. In our model, prototypes and image patches are HOG descriptors rather than natural images. Instead of directly using Euclidean distance, the neural response is computed by a normalized dot-product operation. To evaluate the performance of the proposed model, experiments are conducted on three benchmark databases: Caltech5, Caltech101 and Caltech256. Experimental results show that the proposed HOG-HMAX model not only keeps good tolerance to scale and position change, but also reduces the disturbances of illumination and rotational change. Furthermore, our method has significant advantage in the computational complexity. The recognition performance of HOG-HMAX model has been greatly improved in comparison with the standard HMAX model.

[1]  Nicolas Pinto,et al.  How far can you get with a modern face recognition test set using only simple features? , 2009, 2009 IEEE Conference on Computer Vision and Pattern Recognition.

[2]  Yuan Yan Tang,et al.  Sparse-based neural response for image classification , 2014, Neurocomputing.

[3]  Mohammad Bagher Menhaj,et al.  A robust scene descriptor based on largest singular values for cortex-like mechanisms , 2013, 2013 13th Iranian Conference on Fuzzy Systems (IFSC).

[4]  Zohreh Yaghoubi,et al.  Intermediate layer optimization of HMAX model for face recognition , 2011, 2011 IEEE International Conference on Computer Applications and Industrial Electronics (ICCAIE).

[5]  William Stafford Noble,et al.  Support vector machine , 2013 .

[6]  Weifeng Liu,et al.  Bio-inspired invariant visual feature representation based on K-SVD and SURF algorithms , 2013, ICIMCS '13.

[7]  W. Pitts,et al.  A Logical Calculus of the Ideas Immanent in Nervous Activity (1943) , 2021, Ideas That Created the Future.

[8]  Kun Liu,et al.  Rotation-Invariant HOG Descriptors Using Fourier Analysis in Polar and Spherical Coordinates , 2014, International Journal of Computer Vision.

[9]  Anil K. Jain Data clustering: 50 years beyond K-means , 2010, Pattern Recognit. Lett..

[10]  J. Goo,et al.  Receiver Operating Characteristic (ROC) Curve: Practical Review for Radiologists , 2004, Korean journal of radiology.

[11]  Pietro Perona,et al.  Learning Generative Visual Models from Few Training Examples: An Incremental Bayesian Approach Tested on 101 Object Categories , 2004, 2004 Conference on Computer Vision and Pattern Recognition Workshop.

[12]  Lior Wolf,et al.  Using Biologically Inspired Features for Face Processing , 2007, International Journal of Computer Vision.

[13]  Thomas Serre,et al.  Robust Object Recognition with Cortex-Like Mechanisms , 2007, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[14]  David G. Lowe,et al.  Object recognition from local scale-invariant features , 1999, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[15]  Bill Triggs,et al.  Histograms of oriented gradients for human detection , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[16]  W S McCulloch,et al.  A logical calculus of the ideas immanent in nervous activity , 1990, The Philosophy of Artificial Intelligence.

[17]  T. Poggio,et al.  Hierarchical models of object recognition in cortex , 1999, Nature Neuroscience.