A Novel Equation based Classifier for Detecting Human in Images
This work addresses the challenging task of human detection in computer vision, which is important for applications like surveillance and robotics, but it appears incremental as it builds on shape-based classification methods.
The authors tackled the problem of detecting humans in images by focusing on the head-shoulder shape, developing a new Omega Equation to describe it and designing a classifier based on this equation. The method was tested on various shape datasets and demonstrated satisfactory results, though no concrete numbers were provided.
Shape based classification is one of the most challenging tasks in the field of computer vision. Shapes play a vital role in object recognition. The basic shapes in an image can occur in varying scale, position and orientation. And specially when detecting human, the task becomes more challenging owing to the largely varying size, shape, posture and clothing of human. So, in our work we detect human, based on the head-shoulder shape as it is the most unvarying part of human body. Here, firstly a new and a novel equation named as the Omega Equation that describes the shape of human head-shoulder is developed and based on this equation, a classifier is designed particularly for detecting human presence in a scene. The classifier detects human by analyzing some of the discriminative features of the values of the parameters obtained from the Omega equation. The proposed method has been tested on a variety of shape dataset taking into consideration the complexities of human head-shoulder shape. In all the experiments the proposed method demonstrated satisfactory results.