A Comprehensive Analysis of Moving Object Detection Approaches in Moving Camera
To detect moving objects is a difficult task for various image processing and computer vision applications viz., motion segmentation, object classification and identification, behavior understanding, event detection, object tracking and object locating. T
- PDF / 112,743 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 18 Downloads / 186 Views
Abstract To detect moving objects is a difficult task for various image processing and computer vision applications viz., motion segmentation, object classification and identification, behavior understanding, event detection, object tracking and object locating. The process becomes even more difficult with moving camera as compared to static camera as both camera and object motions are combined in the detection process. Moreover, almost all real time video sequences incorporate pan/tilt/zoom camera which makes it essential to detect objects in moving cameras. This paper presents a survey of various moving object detection techniques in moving cameras and gives insight picture of the methods like background subtraction, optical flow, feature based object detection and blob analysis. It also mentions pros and cons of every technique used so far individually. The use of these approaches and progress made over years has been tracked and elaborated. Keywords Object detection
⋅
Moving camera
⋅
Feature classification
1 Introduction Object detection is defined as the process of detecting a change in the position of the object relative to its surroundings or a change in surroundings relative to the object. Areas that are highly explored in case of object detection incorporate face recognition and pedestrian detection. Object detection has applications in numerous territories of Computer vision, including image retrieval and video surveillance. Some of the applications of moving object detection can be listed as follows: • Detection of pedestrians by moving vehicle having pan/tilt camera; • Detection of obstacles in case of camera mounted on a moving robot; • Detection of moving/flying objects in case camera is mounted on a drone; Neeraj (✉) ⋅ Akashdeep CSE Department, UIET, Panjab University, Chandigarh, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2017 H.S. Saini et al. (eds.), Innovations in Computer Science and Engineering, Lecture Notes in Networks and Systems 8, DOI 10.1007/978-981-10-3818-1_30
277
278
• • • •
Neeraj and Akashdeep
Detection of kids, people and toys in case of hand-held camera; Automated vehicle parking systems; Detection of objects in 360° videos; Detection of moving objects in satellite/space station videos.
Object detection in moving camera is intricate as contrasted to static camera in the sense that both foreground (the moving articles) and the background (rest part other than the items) stay in movement. Camera movement can likewise be characterized in two ways i.e., constrained (where way of moving camera is settled i.e., pan/tilt/zoom) and unconstrained (where the movement of camera is obscure). This study presents a summarization of recent approaches in the field of object detection in moving cameras. There can be numerous methods available that can be applied to do so viz., BS (Background subtraction), optical flow, blob analysis etc. Traditionally, we use object detection in static cameras where background of the video is fixed and only the moving parts are random
Data Loading...
