Download presentation
Presentation is loading. Please wait.
Published byAsher Andrews Modified over 9 years ago
1
1 Lucia Maddalena and Alfredo Petrosino, Senior Member, IEEE A Self-Organizing Approach to Background Subtraction for Visual Surveillance Applications Adviser : Chih-Hung Lin Speaker : Kuan-Ju Chen Date : 2009/04/06
2
2 Author Lucia Maddalena received the Laurea degree (cum laude) in mathematics and the Ph.D. degree in applied mathematics and computer science from the University of Naples Federico II, Naples, Italy. Alfredo Petrosino (SM’02) is an Associate Professor of computer science at the University of Naples Parthenope, Naples, Italy.
3
3 OUTLINE INTRODUCTION 1 METHOD 2 EXPERIMENTAL RESULTS 3 CONCLUSION 4
4
4 1.INTRODUCTION VISUAL surveillance is a very active research area in computer vision The main tasks in visual surveillance systems motion detection object classification Tracking activity understanding semantic description
5
5 1.INTRODUCTION The usual approach to moving object detection is through background subtraction Compared to other approaches, The main problem is its sensitivity to dynamic scene changes light changes moving background cast shadows
6
6 1.INTRODUCTION Background subtraction: Unimodal versus multimodal: Recursive: Pixel-based :
7
7 1.INTRODUCTION Unimodal and multimodal: Basic background models assume that the intensity values of a pixel can be modeled low complexity cannot handle moving backgrounds
8
8 1.INTRODUCTION Recursive recursively update a single background model based on each input frame. Space complexity is lower Background model is carried out for a long time period
9
9 1.INTRODUCTION Pixel-based : assume that the time series of observations is independent at each pixel
10
10 1.INTRODUCTION Our approach is based on the background model automatically generated by a self-organizing method and can be broadly classified as multimodal, recursive, and pixelbased.
11
11 2.METHOD Initial Background Model 1 Subtraction and Update of the Background Model 2
12
12 2.1 Initial Background Model abc d ef a1a1 a2a2 a3a3 a4a4 a5a5 a6a6 a7a7 a8a8 a9a9 b1b1 b2b2 b3b3 b4b4 b5b5 b6b6 b7b7 b8b8 b9b9 c1c1 c2c2 c3c3 c4c4 c5c5 c6c6 c7c7 c8c8 c9c9 d1d1 d2d2 d3d3 d4d4 d5d5 d6d6 d7d7 d8d8 d9d9 e1e1 e2e2 e3e3 e4e4 e5e5 e6e6 e7e7 e8e8 e9e9 f1f1 f2f2 f3f3 f4f4 f5f5 f6f6 f7f7 f8f8 f9f9 Let be HSV components, ex: a 1 =(h,s,v)
13
13 2.2 Subtraction of the Background Model
14
14 2.2 Subtraction of the Background Model Use Euclidean distance to compute C n and C other pixel distance
15
15 2.2 Subtraction of the Background Model
16
16 2.2 Update of the Background Model If best match c m Weight vector A t to update in the neighborhood c m If best match c m isn`t found Not update
17
17 2.2 Update of the Background Model abc d ef a1a1 a2a2 a3a3 a4a4 a5a5 a6a6 a7a7 a8a8 a9a9 b1b1 b2b2 b3b3 b4b4 b5b5 b6b6 b7b7 b8b8 b9b9 c1c1 c2c2 c3c3 c4c4 c5c5 c6c6 c7c7 c8c8 c9c9 d1d1 d2d2 d3d3 d4d4 d5d5 d6d6 d7d7 d8d8 d9d9 e1e1 e2e2 e3e3 e4e4 e5e5 e6e6 e7e7 e8e8 e9e9 f1f1 f2f2 f3f3 f4f4 f5f5 f6f6 f7f7 f8f8 f9f9 If best match c m Computer the weight vector to update background
18
18 2.2 Update of the Background Model a1a1 a2a2 a3a3 a4a4 a5a5 a6a6 a7a7 a8a8 a9a9 b1b1 b2b2 b3b3 b4b4 b5b5 b6b6 b7b7 b8b8 b9b9 c1c1 c2c2 c3c3 c4c4 c5c5 c6c6 c7c7 c8c8 c9c9 d1d1 d2d2 d3d3 d4d4 d5d5 d6d6 d7d7 d8d8 d9d9 e1e1 e2e2 e3e3 e4e4 e5e5 e6e6 e7e7 e8e8 e9e9 f1f1 f2f2 f3f3 f4f4 f5f5 f6f6 f7f7 f8f8 f9f9
19
19 SHADOW DETECTION Foreground Ayalyze Hue-Saturation-Value(HSV) color space shadow mask: define a darkening effect of shadows identifying as shadows those points average image luminance Following three condition to mask shadow
20
20 3.EXPERIMENTAL RESULTS (a) original frame; (b) computed moving object detection mask (c) background model (d) background model change mask from previous frame
21
21 3.EXPERIMENTAL RESULTS (a) original frame; (b) computed moving object detection mask
22
22 3.EXPERIMENTAL RESULTS
23
23 3.EXPERIMENTAL RESULTS (a)test image (b) ground truth (c) SOBS result (d) Pfinder result (e) VSAM result (f) CB result
24
24 4.CONCLUSION This paper also includes a comprehensive accuracy testing, performed with both pixel-based and frame-based metrics Experimental results, using different sets of data and comparing different methods, have demonstrated the effectiveness of the proposed approach illumination changes cast shadows ONGOING WORK improve detection results
25
25 www.themegallery.com
26
26 1.INTRODUCTION Parametric versus nonparametric:
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.