In an effort to help machines see more like people do, two new techniques for computer-vision technology – heat mapping and heat distribution – have been developed to mimic how humans perceive 3-D shapes. The techniques apply mathematical methods to enable machines to perceive 3-D objects by mimicking how humans perceive 3-D shapes and instantly recognizing objects no matter how they are twisted or bent. Building on the basic physics and mathematical equations related to how heat diffuses over surfaces, researchers at Purdue University tested their method on certain complex shapes, including the human form and a centaur. The “heat mean signature” of a human hand model is used to perceive the six segments of the overall shape and to define the fingertips. Courtesy of Karthik Ramani and Yi Fang, Purdue University. Although humans can easily see shapes in three dimensions, it proves more difficult for computers. To combat this problem, the scientists developed a method that accurately simulates how heat flows on the object while also revealing its structure and distinguishing unique points needed for segmentation by computing the heat mean signature. Knowing the heat mean signature allows the computer to determine the center of each segment, to assign a “weight” to specific segments and to define the overall shape of the object. The heat mapping allowed the computer to recognize the objects and to ignore “noise” introduced by imperfect laser scanning and other erroneous data. A new machine-vision technique was tested on complex shapes including the human form and a centaur. The techniques offer many potential applications, including robot vision and navigation; 3-D medical imaging; military drones; a 3-D search engine to find mechanical parts such as automotive components in a database; multimedia gaming; creating and manipulating animated characters in film production; helping 3-D cameras to understand human gestures for interactive games; contributing to progress of areas in science and engineering related to pattern recognition; machine learning; and computer vision. Their findings were detailed in two papers presented June 21-23 at the IEEE Computer Vision and Pattern Recognition conference in Colorado Springs, Colo.