icbv course final project sergey tyrin itamar barkai

33
ICBV Course Final Project Sergey Tyrin Itamar Barkai

Upload: james-joshua-phelps

Post on 18-Jan-2016

217 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: ICBV Course Final Project Sergey Tyrin Itamar Barkai

ICBV Course Final ProjectSergey TyrinItamar Barkai

Page 2: ICBV Course Final Project Sergey Tyrin Itamar Barkai

The main goal of this research is to develop an image analysis system capable of locating

grapes in a 2-dimensional image with no additional

information

Page 3: ICBV Course Final Project Sergey Tyrin Itamar Barkai

1. Shape inference using circular Hough transform

2. Hue and Green colour separation3. Local minima search and Clustering

Page 4: ICBV Course Final Project Sergey Tyrin Itamar Barkai

Circular Hough transform parameters:• The range of radii to search for .• The perfectness threshold of a circle.

Page 5: ICBV Course Final Project Sergey Tyrin Itamar Barkai

Playing with perfectness threshold :

Low threshold High threshold

Page 6: ICBV Course Final Project Sergey Tyrin Itamar Barkai

Combining thresholds :

Threshold combination

Page 7: ICBV Course Final Project Sergey Tyrin Itamar Barkai

Playing with radii threshold :

Too high range of radii Many false detections

Page 8: ICBV Course Final Project Sergey Tyrin Itamar Barkai

All of the vine grape berries have a yellow-greenish hue, which is easily separated by the human vision system from the reddish hue of the trunks and from the almost white colour of the grass in the background.

Page 9: ICBV Course Final Project Sergey Tyrin Itamar Barkai

The leaves have commonly a very similar range of green hue, and a human observer wouldn’t possibly know to separate the grapes from the leaves if he hadn’t had an a-priori knowledge of the shape.

Green hue layer of the RGB

Hue layer of the HSV

Page 10: ICBV Course Final Project Sergey Tyrin Itamar Barkai

Look for more information in other colour spaces separation...

Page 11: ICBV Course Final Project Sergey Tyrin Itamar Barkai

Solution for this was found in the yellow hue layer of the CMYK representation. The sun-highlighted centres of the berries stand out.

Page 12: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 13: ICBV Course Final Project Sergey Tyrin Itamar Barkai

I. Local minima extractionFinding areas of local minima, cutting

of singlepoint minima.

Page 14: ICBV Course Final Project Sergey Tyrin Itamar Barkai

II. Removal of noise detections by spot sizeFiltering out where spot size > 300.95% probability of being part of the

background.

Page 15: ICBV Course Final Project Sergey Tyrin Itamar Barkai

III. ClusteringLooking for condensed groups of spots.

Page 16: ICBV Course Final Project Sergey Tyrin Itamar Barkai

III. ClusteringLooking for condensed groups of

spots.

Page 17: ICBV Course Final Project Sergey Tyrin Itamar Barkai

IV. Removal of edge linesMaking segments more oval by

removing one-pixel-width lines.

Page 18: ICBV Course Final Project Sergey Tyrin Itamar Barkai

V. Filtering of a cluster as one entity•Using RGB, Magenta, and Black colour spaces to calculate median values of clusters as total and comparing these values to each segment median values .•Removing segments where median max value and min value are too far away.•Removing segments where magenta is too high or black is too low (trunk or background).•Applying this filter to the cluster as whole.

Page 19: ICBV Course Final Project Sergey Tyrin Itamar Barkai

VI. Calculation of a hullJoining segments into real clusters by

findingconvex hull – inserts non-grape parts into

clusters.

Page 20: ICBV Course Final Project Sergey Tyrin Itamar Barkai

VI. Calculation of a hullSolution – calculating Delaunay

triangulation and cutting off too long edges.

Page 21: ICBV Course Final Project Sergey Tyrin Itamar Barkai

VII. Generating final maskRemoval of non-grape triangles, hull

filling and final rounding of the segmentation

mask.

Page 22: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 23: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 24: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 25: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 26: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 27: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 28: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 29: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 30: ICBV Course Final Project Sergey Tyrin Itamar Barkai
Page 31: ICBV Course Final Project Sergey Tyrin Itamar Barkai

• Due to the highly saturated and unrestricted lighting conditions, it is impossible to detect each and every grape in the image.

• For normalized lighting conditions, a satisfying result can be achieved.

• For the different lighting conditions, using two different sets of threshold parameters, increases the successful detection rate significantly.

Page 32: ICBV Course Final Project Sergey Tyrin Itamar Barkai

Is grapes detection solved?

Page 33: ICBV Course Final Project Sergey Tyrin Itamar Barkai

• Bright/dark areas – apply noise filtering thresholds to smaller regions.

• Smarter cluster filtering – comprehensive filtering decisions.

• Learning process – clusters and grapes shape, lighting conditions.