Animal Image Detection Based on Generalized Hough Transform
Animal image detection is a hot spot in the field of object detection and has been widely used in
animal husbandry and protection. Due to the large variety of animals, their postures are constantly changing as
the state of motion changes. This uncertainty causes difficulty in detection. In this paper, a detection model
based on generalized Hough transform is proposed for the above problem. The peak point of Hough transform
in the parameter space is surrounded by the secondary peak point, which may lead to missed detection or false
detection. Besides, Hough transform detection algorithm is computationally intensive and the threshold is
difficult to set. Based on the linear and arc detection characteristics, firstly, the image edge pixels are aggregated
into different categories according to the adjacency relationship and then are sequentially stored in the array.
Secondly, the threshold interval is used to detect the line and the arc and the pixels obtained by dynamic
sampling in the array are thresholder. Adaptive estimation is used for linear and arc detection of stored pixel
points according to the threshold interval. Experimental results show that the algorithm can improve the
detection efficiency while reducing the computational cost.