Browse Prior Art Database

A Method of Image Alignment for Thermal and RGB Sensors

IP.com Disclosure Number: IPCOM000246275D
Publication Date: 2016-May-23
Document File: 1 page(s) / 32K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for thermal- Red-Green-Blue (RGB) image alignment.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 100% of the total text.

Page 01 of 1

A Method of Image Alignment for Thermal and RGB Sensors

Thermal sensor technology is going to be pervasive, especially in mobile devices such as smart phones. A Red-Green-Blue (RGB) image can be used for object recognition, but a thermal image provides heat distribution that an RGB sensor cannot. In addition, an RGB sensor can see things beyond the human eye.

One problem is image overlapwhen thermaland RGB images are combined. Overlapping images are misaligned and objects in different depths have various levels of displacement. The image texture between the two is different. Conventional image registration using key features does not work. Temperature distribution is correlated to the object's shape.

A method is needed to minimize the difference between two gradient maps.

The novel contribution is a method for thermal-RGB image alignment.

The implementation steps are:

1. Use thermal and RGB sensors to capture images 2. Extract gradient maps of captured images 3. User selects a region of interest (ROI) or computer vision extracting the ROI (e.g., face region)


4. Minimize the difference of gradient maps and get the alignment distance on ROI


5. Use the extracted distance to align thermal and RGB images

The formula is as follows:

• Et (x,y) is the edge map of thermal image
• E dÞ (x,y) is the edge map of visible image with a displacement d

1