Image Recognition: Military requirements push technology forward

The Strategy Page has an article about how the proliferation of video on the battlefield (e.g. from surveillance cameras and even night vision goggles) is driving technology development. Initially, it was the UK (United Kingdom) that drove image recognition technology with its massive deployment of CCTV in public spaces (e.g. train stations, airports, city centers, and shopping malls). Now, it is the military.

The benefit from computer-assistance in analyzing video comes from the following:

1. Volume (i.e. the sheer number of cameras and, hence, images to be monitored and/or analyzed)
2. Concentration (i.e. the human visual system loses the ability to concentrate effectively on images after about 20 minutes of continuous viewing)
3. Memory (i.e. computers can track and, possibly, predict more things simultaneously than a human can because, generally speaking, computers are not nearly as attention and working-memory limited in the short-term as humans and far outstrip us in their ability to recall video sequences over the long term - they can just play back the video recording off of their hard disks)
   

The article claims that the abilities of computer systems to recognize patterns is rapidly improving and approaching that of humans - that is no small feat as humans are simply amazing in their ability to perform real-time pattern analysis.

One intriguing point made in the article is that the current conflicts are generating a lot of real-world recordings of "bad behavior" that can be used to train and test new pattern analysis and prediction algorithms - training data like this is like gold to us signal processing types!

Enjoy!

Human Auditory System: Artificial Cochlea

A next generation implementation of an artificial cochlea has been developed by researchers at the University of Michigan and Tufts University (both in the USA). The natural cochlea is the sea-shell-shaped structure in the human auditory system that is the final stage in the conversion of air pressure variations (a.k.a. sound waves) into neural impulses that convey the frequency and amplitude information information contained in the sound to the brain for interpretation.

EDN (USA technical magazine) has an article which contains more technical detail and pictures on the advance.

I should point out that this version of the artificial cochlea is not an exact replica in form factor (i.e. shape) or function of a natural cochlea. For instance, the artificial cochlea is planar in shape versus sea-shell-shaped; covers the frequency range of 4200 to 35,000 Hz versus the 20 to 20,000 Hz of the human ear; and isn't continually fine tuned by an external processor to focus on frequency bands of interest, like the human cochlea does when one is, for instance, one's brain is picking out a familiar voice in a cocktail party situation. I mention this not to detract from their excellent work, but instead to give you, curious reader, reasonable expectations as to its specifications and performance.

Evidence: Study of the impact of CSI-style shows on jurors

An overview of the study by researchers at the University of Leicester can be found at Science Daily.

Audio Forensics: New LinkedIn Forensic Audio Group

Phil Manchester (Force Forensic Audio Specialist, West Midlands Police, UK) has just started a Forensic Audio Group on LinkedIn, the popular professional networking web-service. He plans to use it not only as a way to connect people working or interested in this field, but also as a forum for discussions related to it. I have just joined up myself and I encourage you to do the same if you are keen on audio forensics. The basic membership in LinkedIn is free of charge, as is membership in the group.

Testimony: Memory is fallible, study confirms

Cosmos (Australian, popular-science magazine) has an article on research by James Ost, University of Portsmouth (UK) into the accuracy of recall, in this case, of video news reports seen some three months prior. There were a couple of interesting statements that popped out at me:

"Ost said that people who were more likely to come up with false memories scored higher on a scale of "fantasy proneness" than those that did not." (My comment - "fantasy proneness" appears to mean "creative" in this context.)

"Ost said that when DNA testing became available in the U.S. in the early 90s, 80 per cent of death row cases that were exonerated, were found to have been wrongly convicted on the strength of mistaken identity."

Although the article doesn't address it directly, previous studies have shown that accuracy of recall also depends greatly on the amount of time that has past since the event, which gives all the more reason to get witness statements soonest.

Image Forensics: Adobe launches CS4

Adobe has launched the latest version of its Creative Suite, which includes, most importantly for image/video forensic types, Photoshop. Headline features of Photoshop CS4 include:

• GPU-based drawing of documents onscreen (for speed improvements)
  
• 64-bit for Windows (but not for Mac OS-X, at least yet)
  
• Expanded 3-D paint, lighting and rendering tools

Hat tip: Ars Technica

Linguistics: Forensic Linguistics used on text message evidence

Dr Tim Grant, Deputy Director of the Centre for Forensic Linguistics at Aston University (UK), is well known in forensic linguistic circles. ScienceDaily has excerpts from an interview with Dr. Grant following the successful conviction of a murderer that relied partly on Dr. Grant's testimony about text messages allegedly sent by the suspect (since convicted) to throw investigators off the trail. I call your attention to the careful and correct use of the phrases "unlikely" and "more likely" to describe the results of the analysis, instead of a phrase such as "proved conclusively".

In addition, here is a link to a BBC (British Broadcasting Corporation, UK) news story about the case and the analysis of the text message evidence. Very interesting stuff!

Acoustics: Listenging to nature

Using microphones to capture the sounds made by animals in their habitat, minus humans and their man-made sounds. These sounds are known as the "biophony" and are used to sense the quality and character of the animals' habitat. Wired has the story.

Author's comment: It's a good thing that animals don't have privacy rights in the USA (yet!), or these scientists would be in jail for unlawful surveillance activities!

(Via Cross-Spectrum Acoustics)

Fun: Top Ten Amazing Physics Videos

Wired has them. Enjoy!

Acoustics: Science of sound in the sea

Sound requires a medium, such as air, to travel in. Here is a link to a site dedicated to the case where the medium is water. The site belongs to the University of Rhode Island (USA) and it has plenty of information for educators and students about the science of sound and the animals that make underwater sounds, such as dolphins, whales, and fish. I particularly liked the recordings of the animal sounds.

PS. Thanks to C.S. for the link!

Human Auditory System: Perfect pitch is more common than previously believed

It turns out that the ability to identify the pitch of a musical note without needing a reference note to compare to is more common than previously believed. This ability is called "perfect pitch" or "absolute pitch", as opposed to "relative pitch".

Using a testing technique adapted from studies involving infant hearing, researchers at the University of Rochester's Eastman School of Music and Department of Brain and Cognitive Sciences were able to test for perfect pitch in a random sample of people, including people with no musical training whatsoever. In the past, these tests were performed only on people with musical training who could identify notes by name. For more on this study, follow this link.

Fun: Send your voice across a fjord, for the price of a telephone call

Here is a bit of fun - the Telemegaphone.

(Hat tip: Good Morning Silicon Valley)

Human Auditory System: Tinnitus follow-up

FYI, I added some new information to my recent post on tinnitus. It can be found here.

Admin: Posting will be light

Posting will be light for another week or so while I am on summer holidays. Back soon, when I will be "tan, rested, and ready" for more blogging.

Biometrics: Fingerprints versus palm prints

While driving back from Charleston, SC (USA) today - a lovely seaport city, by the way - I was catching up on my podcasts and happened to listen to a Wall Street Journal (US, center-right, business news) segment on new palm print readers.  The part that caught my attention was the claim by one interviewee that - and I'm paraphrasing here - "people" were concerned about using fingerprint biometric identification technologies because they were worried about later being falsely accused of a crime due to a mistaken match to a print left at a crime scene.  The interviewee went on to say that as police did not collect palm prints during crime scene investigations, palm print biometrics systems were getting a leg up in the market place.

This made me wonder exactly who are these "people" he referred to?  I'm rather doubtful that it is the general public that has these concerns.  I find it much easier to believe that elements of criminal "society" or privacy advocates would be tuned-in to these issues, but not the average man on the street.  I might be being too critical of this particular piece, but I thought that the reporter should have challenged, or at least followed up on, that statement.

In a larger context, this may unfortunately be a textbook example of how public policy is determined at the chaotic intersection where the general public, subject matter experts, issue advocates, the marketplace, and technology all collide.  I certainly hope not!