Image-Based Segmentation of Indoor Corridor Floors for a Mobile Robot Yinxiao Li and Stanley T. Birchfield Department of Electrical and Computer Engineering

Image-Based Segmentation of Indoor Corridor Floors for a Mobile Robot

Yinxiao Li and Stanley T. BirchfieldDepartment of Electrical and Computer Engineering

Clemson University{ yinxial, stb }@clemson.edu

MotivationGoal: Segment the floor in a single corridor image

Why?• obstacle avoidance• mapping• autonomous exploration and navigation

MotivationGoal: Segment the floor in a single corridor image

Why?• obstacle avoidance• mapping• autonomous exploration and navigation

Outline

• Previous Work

• Detecting Line Segments

• Score Model for Evaluating Line Segments

– Structure Score

– Bottom Score

– Homogeneous Score

• Experimental Results

• Conclusion

Outline

• Previous Work



– Structure Score

– Bottom Score



• Conclusion

Previous WorkFree space detection

• Combination of color and histogram (Lorigo 1997)

• Optical flow (Stoffler 2000, Santos-Victor 1995)

• Stereo matching (Sabe 2004)

Floor detection• Apply planar homographies to

optical flow vectors (Kim 2009, Zhou 2006)

• Stereo homographies (Fazl-Ersi 2009)

• Geometric reasoning (Lee 2009)

Limitations• Multiple images for motion• Multiple cameras for stereo• Require different colors for floor and wall• Computational efficiency• Calibrated cameras • Assume ceiling visible

Our contribution: Segment the floor in real time using a single image captured by a low-height mobile robot

Challenging problem: Reflections

Also:• variety of poses (vanishing point, ceiling not always visible)• sometimes wall and floor color are nearly the same

Outline

• Previous Work



– Structure Score

– Bottom Score



• Conclusion

Douglas-Peucker Algorithm

Purpose: Reduce the number of points in a curve (polyline)

Algorithm:

1. Connect farthest endpoints2. Repeat

1. Find maximum distance between the original curve and the simplified curve

2. Split curve at this point Until max distance is less than

threshold

David Douglas & Thomas Peucker, "Algorithms for the reduction of the number of points required to represent a digitized line or its caricature", The Canadian Cartographer 10(2), 112–122 (1973)



Algorithm:




threshold



Algorithm:




threshold



Algorithm:




threshold



Algorithm:




threshold



Algorithm:




threshold

Modified Douglas-Peucker Algorithm

original algorithm: dallowed is constantmodified algorithm: dallowed is given by half-sigmoid function

dallowed

original modified

Detecting Line Segments (LS)


1. Compute Canny edges



2. Modified Douglas-Peucker algorithm to detect line segments

Vertical LS: slope within ±5° of vertical direction Horizontal LS: Slope within ±45° of horizontal direction

dallowed



2. Modified Douglas-Peucker algorithm to detect line segments

Vertical LS: slope within ±5° of vertical direction Horizontal LS: Slope within ±45° of horizontal direction

3. Pruning line segments (320x240) Vertical LS: minimum length 60 pixels Horizontal LS: minimum length 15 pixels Vanishing point (height selection): The vanishing point is computed as the mean of the intersection of pairs of non-vertical lines. It is used for throwing away horizontal line segments coming from windows, ceiling lights, etc.

dallowed

Outline

• Previous Work



– Structure Score

– Bottom Score



• Conclusion

Score Model

is assigned to each horizontal line

Structure Score Bottom Score Homogeneous Score

Weightshorizontal line

Score Model – Structure

Given a typical corridor image

1. |I(x,y)| > T1

(How to set threshold T1?)

|I(x,y)| > T1



1. |I(x,y)| > T1 (gradient magnitude)


2. I(x,y) < T2 (image intensity)


|I(x,y)| > T1 and I(x,y) < T2







We applied SVM to 800 points in 200 images:

|I(x,y)| > T1 and I(x,y) < T2







Approximate using two thresholds:

|I(x,y)| > T1 and I(x,y) < T2





3. Now threshold original image using TLC (average gray level of pixels satisfying #1 and #2)

I(x,y) < TLC





3. Now threshold original image using TLC (average gray level of pixels satisfying #1 and #2)

4. Compute the chamfer distance between the line segments and structure blocks


Ridler-Calvard Otsu

Our method

Our method is able to remove the spurious pixels on the floor caused by reflections or shadows

Method R-C Otsu Ours

Correctness 62% 66% 82%

Comparison of different threshold methods

Outline

• Previous Work



– Structure Score

– Bottom Score



• Conclusion

Score Model – Bottom

1. Connect the bottom points of consecutive vertical line segments to create “bottom” wall-floor boundary





2. Compute the distance of each horizontal line segment to the “bottom” wall-floor boundary

(red circled horizontal line segments indicates a positive contribution)

Outline

• Previous Work



– Structure Score

– Bottom Score



• Conclusion

Score Model – HomogeneousIdea: The floor tends to have larger

regions (due to decorations, posters, windows on the wall).

Algorithm:

1. Color-based segmentation of the image (using Felzenszwalb-Huttenlocher’s minimum spanning tree algorithm)

2. For each horizontal line segment, compute size of region

just belowsegment

size of largestsegment

Score Model – HomogeneousIdea: The floor tends to have larger

regions (due to decorations, posters, windows on the wall).

Algorithm:

1. Color-based segmentation of the image (using Felzenszwalb-Huttenlocher’s minimum spanning tree algorithm)

2. For each horizontal line segment, compute size of region

just belowsegment

size of largestsegment

Segmenting the floor

• Normalize the scores and sum• Threshold the final score:

• Connect the remaining line segments and extend the endpoints to the edges of the image

Outline

• Previous Work



– Structure Score

– Bottom Score

– Homogenous Score


• Conclusion

Evaluation Criterion• Wall- floor Boundary

– Green line: Detected wall-floor boundary

– Red line: Ground truth

Evaluation Criterion• Wall- floor Boundary

– Green line: Detected wall-floor boundary

– Red line: Ground truth

• Area– Blue shade area: misclassified area

– Red shade area: ground truth floor area

• Error Rate

• Segmentation is considered successful if rerr < 10%

%err

bluer

red

Sample Results89.1% success on database of 426 images:

Sample ResultsImages downloaded from the internet:

Sample ResultsSome successful results on failure examples from Lee et al. 2008

Original Image

Lee’s Ours

D. C. Lee, M. Hebert, and T. Kanade. “Geometric Reasoning for Single Image Structure Recovery”. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2009.

Sample Results• Failure examples

Checkered floor

Textured wall

Darkimage Bright

lights

Sample Results

• Video clip 1

Sample Results

• Video clip 2

Outline

• Previous Work



– Structure Score

– Bottom Score



• Conclusion

Conclusion• Summary

- Edge-based approach to segmenting floors in corridors

- Correctly handles specular reflections on floor- Nearly 90% of the corridor images in our database

can be correctly detected.- Speed: approximately 7 frames / sec

• Future work- Speed up the current algorithm- Improving score model by add more visual cues

(highly textured floors, low resolution, or dark environment)

- Use floor segmentation for mapping

Acknowledgement Clemson Computer Vision Group reviewers

• Zhichao Chen• Vidya Murali

Anonymous reviewers

Thanks!

Questions?

Image-Based Segmentation of Indoor Corridor Floors for a Mobile Robot

Documents

Image-Based Segmentation of Indoor Corridor Floors for a Mobile Robot Yinxiao Li and Stanley T. Birchfield Department of Electrical and Computer Engineering