BROWSE
Detail
Conference Paper
Object Detection With Sliding Window in Images Including Multiple Similar Objects
Cited 35 time in 
Share 
Share
- Authors
- Jinsu Lee, Junseong Bang, Seong-Il Yang
- Issue Date
- 2017-10
- Citation
- International Conference on Information and Communication Technology Convergence (ICTC) 2017, pp.804-807
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- Given an image containing an object of interest, the object can be detected by comparing the feature points in the given image with those in a reference object image. In a case where the given image contains a large number of similar objects, the object of interest is difficult to be detected. It is because the feature points in the given image are concentrated in some regions where each region has a drastic change in the intensity of points. One of the methods to overcome the problem of the feature concentrating is to increase the limitation in the number of feature points to be used for the detection. However, this causes more computational load. Alternatively, the resolution of the image can be lowered, but this method decreases the accuracy of detection. In this paper, in order to detect the object of interest in an image with multiple similar objects, a sliding window for feature matching is used. The sliding window is optimized in size for better performance of the object detection. As a practical example, a service that visualizes the location of a desired book in a library is considered. The image for feature matching is obtained from high-resolution CCTVs which are connected to a cloud server that has databases of book title images. This service can be extended to visualize the location of the object to the users, using augmented reality (AR) technology on a mobile platform of a smart phone or smart glasses. The presented method for this service is experimentally evaluated.
- KSP Keywords
- Accuracy of detection, Augmented reality(AR), Cloud server, Feature matching, High-resolution, Mobile platform, Object detection, Object image, Sliding Window, Smart Phone, computational load