Browse Prior Art Database

TRACKING METHOD USING INFRARED PICTURES

IP.com Disclosure Number: IPCOM000146207D
Publication Date: 2007-Feb-07
Document File: 2 page(s) / 137K

Publishing Venue

The IP.com Prior Art Database

Related People

Pierrick Jouet: INVENTOR [+3]

Abstract

BACKGROUND In order to track an object in a video, state of art methods use a search window. This window is often initialized by a user on the first picture of the video and then positioned along the video by a tracking algorithm. The object itself is then segmented on each picture of the video by delimitating the segmentation to the search window. CONCEPT The following method is proposed to improve object tracking. In one embodiment, thermal pictures and traditional color pictures are used. Initially, the thermal pictures and the color pictures are registered (step 1). Subsequently, a thresholding step is applied to the thermal pictures (step 2) to obtain binary masks identifying an object to be tracked. Finally, the intersection of the search window positioned along the video and of the binary mask is used (step 3) to delimitate the region for segmenting the object. The steps are described in more detail below. Picture registration (step 1) To simplify and as depicted in Figure 1, it is assumed that the camera axes, focal and zoom values are known. Furthermore it is assumed that both cameras are in the same plane. Figure 1 Thresholding the thermal picture (step 2) In one embodiment, for a thermal picture in white and black, the grey levels correspond to the temperature gradients. If it is assumed that the temperature of an object is homogeneous, applying a thresholding step identifying a low threshold LT and a high threshold HT makes it possible to identify a given object precisely. Figure 2 For example, in Figure 2 above, a thresholding step with values of LT=230 and HT=250 makes it possible to identify the head and arms of the man. With values of LT=150 and HT=250, it is possible identify the man entirely. tracking(step 3) In a tracking step, at least one search window is located in a first picture. It should be noted that the location of the search window(s) can be performed automatically or manually. The search window(s) is tracked along the sequence and object segmentation is achieved in this window. Step 3 consists of validating a search window located in a current image before segmenting the object. If the intersection of the search window and of the binary mask is not empty, the window is validated, otherwise a tracking error is detected.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 59% of the total text.

Tracking method using infrared pictures

BACKGROUND

          In order to track an object in a video, state of art methods use a search window. This window is often initialized by a user on the first picture of the video and then positioned along the video by a tracking algorithm. The object itself is then segmented on each picture of the video by delimitating the segmentation to the search window.

AUTHORS

Pierrick Jouet

François Le Clerc

Thierry Tapie

CONCEPT

The following method is proposed to improve object tracking. In one embodiment, thermal pictures and traditional color pictures are used. Initially, the thermal pictures and the color pictures are registered (step 1). Subsequently, a thresholding step is applied to the thermal pictures (step 2) to obtain binary masks identifying an object to be tracked. Finally, the intersection of the search window positioned along the video and of the binary mask is used (step 3) to delimitate the region for segmenting the object.  The steps are described in more detail below. 

Picture registration (step 1)

To simplify and as depicted in Figure 1, it is assumed that the camera axes, focal and zoom values are known. Furthermore it is assumed that both cameras are in the same plane.

 

                        Figure 1

Thresholding the thermal picture (step 2)

In one em...