Semi automatic target initialization method based on visual saliency

• US 9,595,114 B2
• Filed: 08/28/2013
• Issued: 03/14/2017
• Est. Priority Date: 08/28/2013
• Status: Active Grant

• Current Assignee: Aselsan Elektronik Sanayi Ve Ticaret Anonim Sirketi
• Original Assignee: Aselsan Elektronik Sanayi Ve Ticaret Anonim Sirketi
• Inventors: Tunali, Emre, Oz, Sinan, Aytekin, Caglar
• Primary Examiner(s): Hung, Yubin
• Application Number:
  US14/915,627
• Publication Number:
  US 20160239981A1
• Time in Patent Office: 1,294 Days
• US Class Current: 1/1
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 7/11 Region-based segmentation

Abstract

Target initialization can dramatically change the performance of the tracker, since the initial window determines for the tracker what to track. In order to achieve a better tracking performance; The present invention relates to a method of semi automatic target initialization based on visual saliency for a given point coordinate in the vicinity of target by the user. Performance boost of tracker is mainly based on two key features of target initialization algorithm: It is capable of compensating erroneous user input; also selecting the most distinctive, salient part of object as target, so better discrimination is achieved between the target and background. Experimental results show that tracking performance is boosted in scenarios, in which the tracking is initialized by the proposed algorithm. Very low computational cost and requirement of only a point coordinate as input in the neighborhood of the target make this approach preferable in real time tracking applications.

Claims

1. A method for semi automatic target initialization (100) comprises the steps of,

inputting an image coordinate by user (101),

selecting an initial window around the image coordinate by center-surround histogram distance (102),

generating saliency map of the window (103),

binarizing the saliency map (104),

selecting a target (105),

outputting a bounding box enclosing the target (106) and characterized by the following steps for binarizing the saliency map;

detecting local maxima of the saliency map and sorting them in descending order to obtain a vector LocalMax_sorted (402),

calculating a normalized laplacian of the vector with the formula;

2. A method for semi automatic target initialization (100) according to claim 1, comprising the steps of;

obtaining a foreground histogram (202),

obtaining a background histogram (203),

calculating a quadratic-chi histogram distance (204),

obtaining a histogram distance vector by repeating steps 202,203 and 204 for each window size (205),

selecting 1st local maximum of the histogram distance vector as initial window size (206) in the step “

selecting an initial window via center-surround histogram difference (102)”

.

3. A method for semi automatic target initialization (100) according to claim 1, comprising the steps of,

calculating a mean intensity for patches obtained by any super pixel selection algorithm (302),

calculating a mean intensity difference in eight neighborhood and assign as weights (303),

calculating the shortest path from each foreground patch to a background patch (304),

returning the saliency map by normalizing the values calculated as the shortest path (305).

4. A method for semi automatic target initialization (100)) according to claim 1, comprising the steps of,

executing connected component analysis on the threshold saliency map (501),

inputting each component to the energy function and selecting the component maximizing the regularization energy as target (502),

reporting the bounding box of the connected component as target having the most salient region with minimum distance to center with formula;

5. A method for semi automatic target initialization (100) according to claim 1, wherein the step “

initial window selection via center-surround histogram difference (102)”

defines proper window that includes foreground and background which is capable of fulfilling the assumptions that are made in saliency generation method.

6. A method for semi automatic target initialization (100) according to claim 1, wherein initial window selection is started with histogram distance vector generation and finalized by selection of initial window size which corresponds 1^st local maximum of the histogram distance vector.

7. A method for semi automatic target initialization (100) according to claim 1, is suitable for long-term tracking since it boosts performance of tracker by achieving better discrimination between the target and background.

8. A method for semi automatic target initialization (100) according to claim 1, wherein target bounding box is found with just a click by the user in step “

image coordinate given by user (101)”

turns the target initializer into a user friendly framework.