A novel algebraic method for kernel-based object tracking☆
Graphical abstract
Introduction
Due to the simplicity and efficiency of mean-shift tracking, it is a common method for kernel object tracking. In the original mean-shift tracking, a symmetric weighted histogram kernel and a geometric shape are used for modeling the target [1]. The motion of the kernel from one video frame to the next is computed by iteratively estimating the color centeriod for reach to the best similarity between the candidate and the target models. Including the spatial information, predictive component, radial basis function network and multiple reference histograms to the mean-shift tracking are proposed for better performance [2], [3], [4], [5], [6]. Many efforts are made to solve the original mean-shift drawbacks related to its scale and orientation changes, symmetric kernel and isotropic shape. Because of change in object size, the fixed scale model is not appropriate for tracking. Using three different kernel bandwidths (scales) for algorithm and selecting the one which maximizes the appearance similarity is proposed by Comaniciu [7]. This method, named as ±10% method, is computationally expensive due to the brute-force nature. Using difference of Gaussian mean-shift kernel in scale to include scale as additional dimension has been proposed in [8]. This method performs well; but, it needs convolving the image with a set of Gaussian filters at various scales which is a computationally complex task. Using asymmetric kernel and considering the scale and orientation as additional dimensions to the spatial image in which the tracking algorithm is solved simultaneously in all coordinates can be seen in recent works [9], [10], [11], [12]. Paying attention to the moment features of the weight image of the target candidate region and the bhattachryya coefficients and also using the principal components of the variance matrix of the spatial-color distribution are proposed to solve the scale and orientation problems [13], [14]. Another limitation is related to the geometric region and its asymmetric kernel. In cases where the object does not have an isotropic shape, the symmetric kernel which is isotropic in shape cannot describe the object properly, and it can be contain the considerable background information. Using a Shape-adapted asymmetric kernel constructed based on the mask of the detected object can improve the tracking result [15], [16], [17]. Despite various efforts to improve the mean-shift algorithm, there is an intrinsic problem in all mean-shift methods. The problem is related to the iterative process of mean-shift tracker for finding the best similarity between the candidate and the target model. In the first iteration, the candidate and model regions are considerably inconsistent. Thus, the influence of background information can be significant, especially in objects which are undergoing rapid motions or are spatially adjacent objects. Of course, this is the problem of every tracker which finds the target by searching for the best match such as contour [18], [19], [20], [21] and template tracking [22], [23].
In this paper, we propose a new method for tracking the target area in which the displacement of the kernel is obtained using the previous object region as a starting point in the current frame. In contrast with the mean-shift tracking which finds the kernel location using an iterative technique to achieve the best histogram matching, the proposed method offers algebraic equations for the kernel displacement. Expressing the displacement kernel using algebraic equations is much more effective than the mean-shift way to remove the background. In order to eliminate the effect of background, we have defined two types of disconnected and connected background. Defining disconnected background allows the algorithm to consider the object and background areas as individual objects. We acquire the corresponding target area for each of these objects using tracking equations. The object and the background can be identified with the weighted histogram, which is defined as a metric to measure the similarity. The background which is connected to the object can be eliminated using the segmentation of object and also modified histogram. The experiments results depict the validity of the proposed tracking equations and its advantages over the mean-shift method to remove the effect of the background.
The paper is organized as follows. In Section 2, the target model is presented. Section 3 describes the proposed method for calculating the displacement of kernel and removing the background effect. Experimental results and conclusion are given in Sections 4 Results and discussion, 5 Conclusion.
Section snippets
Target model
We define the target model based on the binary mask in the first frame and also histogram and window that are associated with the mask region. If (xk, yk) denotes the kth pixel position in R2 (image) space, the binary mask object locations with n pixels (which obtained from previous object) can be defined as:
The color histogram feature of the target model for binary mask with color u (u = 1, …, m) is considered as:where δ is the kroneker
Proposed method
The proposed method uses a novel idea which relates the displacement of object to the previous object center and the center of the target region which lies in the previous object region. The relation is in the algebraic form which can be used to remove effectively the background information. We describe this process in detail in the following subsections, which is organized as follows. Section 3.1 describes how to obtain the object pixels of the current frame which are in the previous area. The
Results and discussion
The proposed algorithm is tested on several video sequences which include single object indoors, single object outdoors and multiple objects with shape variation and scale changes in both static and cluttered background. The region, geometric center and bounding box of the object were initialized in the first frame. The proposed tracking algorithm is initialized by segmenting the object and computing the histograms of segments. The histograms remain constant during the course of tracking. The
Conclusion
This paper has demonstrated a new kernel-based object tracking. The previous object region and pixels of the target in this region are used for extracting the movement of kernel. The tracking relations are in algebraic form which allows division of the target area in the previous object into two parts and detection of the object and background regions. We propose to divide the background into two types of connected and disconnected regions. The connected background effect which is due to
Fayzollah Khakpoor received the B.S degree in Electronic from Electrical and Electronic Engineering department of Amir Kabir University, Tehran, Iran, in 1989 and his M.Sc. degree from Tehran University in 1996. He is currently Ph.D. student in department of Electrical, Electronic and System Engineering of Noushirvani University, Babol, Iran.
References (24)
- et al.
A spatial-color mean-shift object tracking algorithm with scale and orientation estimation
Pattern Recogn Lett
(2008) - Comaniciu D, Ramesh V, Meer P. Real-time tracking of non-rigid objects using mean shift. In: IEEE conference on...
- Birchfield S, Rangarajan S. Spatiograms versus histograms for region-based tracking. CVPR, vol. 2; 2005. p....
- Fan Z, Yang M, Wu Y. Multiple collaborative kernel tracking. In: PAMI, vol. 29(7); 2007. p....
- et al.
Mean shift tracking with multiple reference color histograms
Image Vis Comput
(2009) - et al.
Online adaptive radial basis function networks for robust object tracking
Comput Vis Image Underst
(2009) - et al.
Applying mean shift motion information and Kalman filtering approaches to object tracking
ISA Trans
(2012) An algorithm for data-driven bandwidth selection
IEEE Trans Pattern Anal Mach Intell
(2003)- Collins R. Mean-shift blob tracking through scale space. In: IEEE conference on computer vision and pattern...
- Yilmaz A. Object tracking by asymmetric kernel mean shift with automatic scale and orientation selection. In: IEEE...
Cited by (2)
Multi-technique object tracking approach- A reinforcement paradigm
2018, Computers and Electrical EngineeringCitation Excerpt :This method basically track objects by finding the most similar distributed pattern in a sequence of frames by iterative searching [11]. However, its major setback is that it fails when the target object moves toward or away from the camera's focus [7,12]. To address this problem, the continues adaptive mean-shift (CamShift) which adaptively adjusts the tracking window's size and the distribution pattern of the target object during tracking was proposed [13].
Fuzzy grey prediction-based particle filter for object tracking
2016, Mathematical and Computational Applications
Fayzollah Khakpoor received the B.S degree in Electronic from Electrical and Electronic Engineering department of Amir Kabir University, Tehran, Iran, in 1989 and his M.Sc. degree from Tehran University in 1996. He is currently Ph.D. student in department of Electrical, Electronic and System Engineering of Noushirvani University, Babol, Iran.
Gholamreza Ardeshir received his B.Sc. from Ferdosi University (Mashhad-Iran) in 1989, M.Sc. from Tarbiat Modares University (Tehran-Iran) in 1993 and Ph.D from University of surry (Guildford-UK) in 2003, respectively. Since 1994, he has been a member of Electrical Engineering Faculty at Babol University of Technology.
- ☆
Reviews processed and recommended for publication to Editor-in-Chief by Deputy Editor Dr. Ferat Sahin.