Homogeneity Property Based Two Wheeler Detection Research Using Histogram of Oriented Gradients and Soft Computing

  • Authors

    • Yeunghak Lee
    • Israfil Ansari
    • Jaechang Shim
    https://doi.org/10.14419/ijet.v7i4.36.29010
  • Object detection, Histogram of oriented gradients, Homogeneity, Adaboost, Two-wheelers.
  • We describe an algorithm to detect two-wheelers with various shapes and viewpoints using the homogeneity property and Histograms of Gradients (HOG). The typical shape of a two-wheeler is can be divided into two parts: human body (upper area) and assembled complex components (bottom). The upper area is substantially homogeneous, because the front and rear body views are simple, whereas the bottom part consists of complex shapes and shows little homogeneity. And the bottom area has very little homogeneity, because it is consisted of complicated shapes. Our algorithm using HOG features based on homogeneity to the upper and lower parts separately and uses Adaboost as the classifier. Furthermore, this paper applied the Adaboost algorithm to classify the objects as the soft computing. Our improved algorithm correctly identified more than 84% of the hardest case tested – motorcycles leaning at 60º - and more than 98% of all types of two-wheelers at 90º.

     

     
  • References

    1. [1] Jung H, Ehara Y, Tan JK, Kim H, and Ishikawa S, Applying MSC-HOG Feature to the Detection of Human on a Bicycle, International Conference on Control, (2012), pp: 514-517.

      [2] Cho H, Rybski PE, and Zhang W, Vision-based Bicyclist Detection and Tracking for Intelligent Vehicles, Intelligent Vehicle Symposium, (2010), pp: 454-461.

      [3] Yu L, Zhao F, and An Z, Locally Assembled Binary Feature with Feed-forward Cascade for Pedestrian Detection in Intelligent Vehicle, Int. Conf. on Cognitive Informatics, (2010), pp: 458-463.

      [4] Zhu Q, Yeh MC, Cheng TK, and Avidan S, Fast human detection using a cascade of histograms of oriented gradients, IEEE Conference on Computer Vision and Pattern Recognition, (2006), pp: 1491-1498.

      [5] Watanabe T, Ito S, and Yokoi K, Co-occurrence Histogram of Oriented Gradients for Detection, Advances in Image and Video Technology(LNCS), (2009), pp. 37-47.

      [6] Wang XY, Han TX, and Yan S, An HOG-LBP human detector with partial occlusion handling, International Conference on Computer Vision, (2009), pp:32-39..

      [7] Mikolajczyk K, Schmid C, and Zisserman A, Human detection based on a probabilistic assembly of robust part detectors, in Proc. Eur.Conf. Comput. Vis., (2004), pp: 69-81.

      [8] Dalal N and Triggs B, Histogram of Oriented Gradients for Human Detection, IEEE Conference on Computer Vision Pattern Recognition, (2005), pp: 886-893.

      [9] Homogeneity, https://en.wikipedia.org/wiki/Homogeneity_and_het- erogeneity#Homogenity (accessed September 2018).

      [10] Viola P, Jones M, and Snow M, Detection pedestrians using patterns of motion and appearance, ICCV 2003, (2003), pp. 153-161.

      [11] Receiver operating characteristic, http://en.wikipedia.org/wiki/Rec-

      eiver_operating_characteristic (accessed September 2018).

  • Downloads

  • How to Cite

    Lee, Y., Ansari, I., & Shim, J. (2018). Homogeneity Property Based Two Wheeler Detection Research Using Histogram of Oriented Gradients and Soft Computing. International Journal of Engineering & Technology, 7(4.36), 1472-1475. https://doi.org/10.14419/ijet.v7i4.36.29010