Human eye-inspired solution provides 10 times better vision to computers

WASHINGTON - Fascinated by the working of the human eye, computer scientists at Boston College (BC) have developed a new technique that allows computers to see objects as fleeting as a butterfly or tropical fish with almost double the accuracy and 10 times the speed of earlier methods.

The novel linear solution has direct applications in the fields of action and object recognition, surveillance, wide-base stereomicroscopy and three-dimensional shape reconstruction.

In the research, Hao Jiang and Stella X. Yu developed a novel solution of linear algorithms to streamline the computer’s work.

Previously, computer visualization relied on software that captured the live image, then hunted through millions of possible object configurations to find a match.

Thus, instead of combing through the image bank, which is a time- and memory-consuming computing task, the researchers in the new study turned to the mechanics of the human eye to give computers better vision.

“When the human eye searches for an object it looks globally for the rough location, size and orientation of the object. Then it zeros in on the details. Our method behaves in a similar fashion, using a linear approximation to explore the search space globally and quickly; then it works to identify the moving object by frequently updating trust search regions,” said Jiang.

Trust search regions act as visual touchstones, to which the computer returns again and again.

The new solution focuses on the mathematically generated template of an image, which looks like a constellation when lines are drawn to connect the stars.

Using the new algorithms, computer software identifies an object using the template of a trust search region.

Then the program adjusts the trust search regions as the object moves and finds its mathematical matches, relaying that shifting image to a memory bank or a computer screen to record or display the object.

Jiang has said that using linear approximation in a sequence of trust regions enables the new program to maintain spatial consistency as an object moves and reduces the number of variables that need to be optimised from several million to just a few hundred.

And thus the speed of image matching increased to 10 times compared with previous methods.

After testing the software on a variety of images and videos - from a butterfly to a stuffed Teddy Bear -a 95 percent detection rate was achieved at a fraction of the complexity.

Jiang said that the so-called “greedy” methods of search and match in the past achieved a detection rate of approximately 50 percent.

The study will be presented at the upcoming annual IEEE meeting on computer vision. (ANI)


Leave a Reply

Protected by Comment Guard Pro