Video analytics textbook released

 - 
Monday, February 15, 2010

BOCA RATON, Fla.—As definitions and nomenclature continue to evolve in the field of video analytics, two editors have taken a stab at codifying things with a new textbook from CRC Press, Intelligent Video Surveillance: Systems and Technology.

“This would be good as a fundamental textbook,” said Gang Qian, an assistant professor with the School of Arts, Media and Engineering and the School of Electrical, Computer and Energy Engineering at Arizona State University, Tempe. A reader, he said, “will know what’s going on in the field, what are the current challenging problems in the areas of video surveillance, and get an idea of the techniques and the mathematical tools that are used in this area.”

“We had two intended audiences,” said Yunqian Ma, a principal research scientist in Honeywell Labs and a senior member of the IEEE. “The first is in the industry, the second in the academic schools, the graduate students who are interested in practicing in this area. As we mentioned in the book, for practitioners this will provide the practical implementations for those who are new to video surveillance. For graduate students and researchers, this will give them a handy reference to summarize the state of the art in video surveillance.”

The book is organized into five parts: The first portion provides an overview of the field, the second and third parts are low-level video surveillance learning modules. The fourth part gets into system design, and the book finishes with real-world applications.

Both researchers readily acknowledge the confusion that surrounds video analytics and the lack of clarity regarding application appropriateness. They hope to dispel some of that confusion.

For example, what is video analytics? “It’s something that extracts useful information from the video,” said Qian. “Maybe that’s as simple as pixel count, maybe that’s useful in some scenarios, but maybe you need something more complicated in other scenarios. To us, the line is not high, it’s just extracting useful information.”

What are the applications where video analytics are consistently a good fit? There, they are less confident.

“It’s hard to say,” said Qian. “It really depends on the application. In one particular environment, the detection tracking can be used well, but the same technology might not work as well in another environment, if there’s changing weather conditions or too many objects in the field. The same technology that works well in one scenario cannot be transferred to another scenario. The complication of real-world scenarios makes it difficult to make any blanket statements.”

“There may need to be another book,” Ma joked, “that pays attention to practical things.”