Archive for the ‘Photography’ Category

Developing Rules of Thumb in Video and Photographic Analysis

Sep 10th, 2015 - Filed under: Blog,Image Analysis,Photography,Video Analysis

A common request for those of us who work in video analysis (whether privately or for a law enforcement or government agency) is to enhance a license plate or compare the person (face, article of clothing, etc.) in a video to a photograph of a person. When is there adequate information to expect that various methods of enhancement may generate value? I develop rules of thumb for these situations.

I think it makes sense for analysts to experiment with files to develop their own rules of thumb as well. In this post I’ll share some of the methods I’ve used to develop my own rules.

License plates. The first thing I did regarding license plates was to determine how many digits may be in a plate, and what the average, minimum number of pixels required are to display the plate. If the state or country uses seven digits, and we consider that most digits require a minimum of three pixels horizontally and one pixel between digits, then a minimum of 27 pixels can, theoretically display most letter and number combinations. If the plate has a couple of W’s and M’s then 27 may not be enough, and if it has fewer digits, or if all the digits are the Letter ‘I’ then fewer pixels would suffice. So, that’s a starting point, but there is more to this.

Do real world examples match this theoretical approach, and are there other things to look for that may be of value? How do compression, lighting, and recorded tonal values affect this? This is where one needs to experiment. You can either take high quality photographs of license plates, or download them from the internet. Then, make multiple versions of these at a variety of pixel resolutions, at a variety of compression settings, at varying degrees of reduced contrast. Save these files and begin analyzing them (starting with the worst) to determine what observations you can make at what settings, sizes, etc. For instance, can you determine the colors in the plate at 20 pixels wide? What resolution enables you to determine the difference between a dealer plate and a standard plate? How many tonal values are required before you can distinguish various characters? As you do this analysis, take notes and see if you find consistency from one plate to another. Start to make other observations – such as when can you differentiate between a ‘5’ and an ‘S’ or a ‘1’ and an ‘I’?

Follow this same approach for other items. For faces, I do a few things. I download mug shots of celebrities, save photos of myself, family, colleagues, and friends, and download random photographs of strangers. As with the license plates, I save multiple copies at varying resolutions, degrees of contrast, compression settings, etc. Then I make my observations. How small of an image shows me an accurate rendition of hair and skin color? When can I determine eye shape or nose shape? When do blemishes begin to show? How about tattoos? How much contrast is needed for a tattoo or blemish to show?

By doing these experiments, we can each have a fairly good idea of what level of resolution, quality of lighting, degree of compression, etc. is necessary to get what level of value from the images we work with daily. We can have some rules of thumb, and we can provide realistic expectations to our colleagues or clients when they bring us images. We can easily explain not only what those realistic expectations might be, but also why, and show them examples to help them understand what limitations may exist.

Raw vs. TIFF for Digital Camera Capture

Jun 18th, 2015 - Filed under: Blog,Fingerprints,Photography

Someone from a major law enforcement agency recently asked me for my opinion regarding the best choice between capturing raw or TIFF files from a digital camera. His concern was that the raw workflow is slower, results in duplicate images (the raw original files and the working TIFF files), and that the raw files are proprietary, and therefore different for each camera brand. Please note that this is only about digital camera capture – not about captures from flatbed scanners.

His concerns are all understandable, but I believe there is a simple answer to each of them.

Is the raw workflow really slower? If a raw file is opened using Adobe Camera Raw and adjustments are made to the image, then opened, it is then in Photoshop and ready for additional processing. With a TIFF file, one skips using Camera Raw, but the time spent processing either file is about the same. But, you can also synchronize the Camera Raw adjustments across multiple raw files with a single mouse click, and then use the Image Processor to convert these to TIFF files for working copies. This gives the speed edge to the raw files.

This leads to the issue of creating duplicate files. But, is this a real issue? The original file that is archived needs to remain unchanged – regardless of file format. When working on that image, the analyst may be making adjustment layers, applying smart filters, sizing the image, etc. That file will be saved as a working file – creating a ‘duplicate’ regardless of the original file format. Since the original raw is significantly smaller (requires less storage space) and contains substantially more information than the camera-generated TIFF, the win goes to the raw file again.

Raw files are proprietary. This is true, and has caused problems when camera manufacturers have stopped supporting raw file formats in the past. The solution to this is the DNG file format. This format is a container that can embed the raw file as well as the information needed to open it, thus providing a more universal file format that will last into the future. DNG files can be automatically generated as the files are downloaded from the camera. TIFF files are also fairly universal – although, if they contain proprietary data, such as Photoshop Adjustment Layers or Smart Filters, then they can only be opened by Photoshop, for example, thus making them proprietary as well.

Raw files are also considered to be a ‘digital negative.’ This is because one cannot make adjustments to the raw file and save the adjusted raw file. Instead, the adjustments made to raw files are sets of instructions that exist either as a separate, XML file, or within a database. When the raw file is opened, those adjustments can be disabled and the original image viewed as captured. Although following best practices with any file format will preserve the original captured image, the raw file format is like putting an extra deadbolt (or ten) on the file.

What about image quality? A raw file is high-bit (usually 12 or 14 bits of information) but a camera-generated TIFF file will be automatically converted by the camera (compressed, downsampled) to 8 bits. The raw file has no embedded color balance or color space, but the TIFF has both. These three things (smaller bit depth, embedded color balance, and embedded color space) make it so that the TIFF will lose quality more quickly with image adjustments than the raw file. And, the raw file will contain more tonal values by orders of magnitude. The camera-generated TIFF image is much more like a camera processed JPEG than a raw file. A strong advantage goes to the raw file.

Next, let’s consider the SWGFAST recommendation regarding images of fingerprints being captured and preserved in a lossless file format. The intent of this recommendation is to enable us to get the best capture possible, and preserve the most information, from our equipment, but I think it leads to some misunderstanding.

A TIFF image, that is generated in-camera, is uncompressed, so it meets this recommendation – but it is actually much more like a camera-generated JPEG than like a raw file. There is quite a bit of in-camera processing that has taken place to generate that TIFF as mentioned above. A large amount of the image data is thrown away by the camera when creating the camera-generated TIFF file – these same processes occur when the camera generates a JPEG image! A camera-generated JPEG image at the camera’s highest quality setting will be virtually indistinguishable from a camera-generated TIFF file – except that the TIFF file will be substantially larger.

Why is an 8-bit TIFF file larger than a 12 or 14 bit raw file? The digital camera doesn’t actually record color information. Instead, it records brightness values for each pixel position. The CCD or CMOS chip is covered with red, green, and blue filters, with each filter position following a specific pattern. The raw file simply records the brightness value of each pixel, and later applies the color interpolation only when the image is opened with a raw file converter, such as Adobe Camera Raw. But, the TIFF file has already applied this color interpolation, so each pixel needs to be described as three separate values, or channels. To measure the number of bits needed to store an uncompressed file, just multiply the pixel width by height, times the bit depth, times the number of channels. Although the bit depth of the TIFF file is less, the number of channels is tripled, making an uncompressed TIFF about twice the size as an uncompressed raw file, even though the raw file contains much more information, and will hold up to a significantly higher amount of image processing.

This leads to the question of whether a JPEG file is really as bad as we have been led to believe. I’ll address that issue in a future post – but I personally prefer a raw workflow and think that the benefits of shooting raw are significant, and that there is no real cost.

Forensic Photographic Analysis – Reverse Projection

Oct 25th, 2013 - Filed under: Blog,Expert Witness,Image Analysis,Photography

Some cases are unique – and this will describe one of those. In this case plaintiff retained me to determine if a photograph of a part was broken. The part was a water jet from a bathtub spa. The person using the spa was injured, and the reason proffered was that one of the legs that hold the jet in place was broken. The hotel personnel recovered the broken jet, placed it on a table and took one photograph of it. Later, the jet was lost before any expert could analyze it.

I was retained to determine if the leg of the water jet in the photograph was broken. The jet had three legs and the one that was believed to be broken was the one on the back side when looking at the photograph. Looking at the photograph, that leg appeared to be at a slightly different angle, but defense claimed that the leg was not broken and that the slight difference in angle may have been from the angle of the photograph, the lighting, the focal length of the camera lens, or simply the way that the jet was positioned on the table – or perhaps it wasn’t at a slightly different angle at all.

Measuring the angles of the legs to the front of the jet wasn’t possible because we had only one image and the leg in question was on the back side of the jet. A 3-D model couldn’t be built because there was only one photograph. But, a reverse projection analysis could be done. Reverse projection involves taking a new photograph from the exact same position as the original image, then one can have the data needed to make comparisons, make measurements, compare lighting, etc.

In this case, plaintiff provided me with a new, unbroken, water jet of the same make and model as in the original photograph. I placed it on a table in the exact same position as in the original. I set up lights to exactly replicate the lighting in the original, with all highlights and shadows matching. I set my camera into the same position that was used in the original photograph, and I took the picture. Next, I created an overlay of the two photographs. The result was conclusive – everything on the two water jets matched except for the one leg. The leg had to be broken on the jet in the original photograph, as there was no other explanation for the difference in the two photographs. The difference between the angle of the leg in the original photograph and in the new photograph were measured, and made part of the illustration.


Countering Bad Science

When attorneys get a copy of opposing counsel’s expert reports, sometimes they are quite surprised by what they read – it’s just bad science. And, in these cases they will then retain their own expert to counter, and possibly quash, the other experts opinion. I’ve been retained in numerous cases to do this, and I’ll discuss four to illustrate the benefit this has, and to help illustrate the kind of work that I do.

Recently, the Los Angeles District Attorney’s Office retained me because they suspected that the forensic video analysis and photogrammetry performed by the defense wasn’t based on science. I was provided with the security video from the business where the crime occurred, and with the report by the defense expert. In this case, the defense expert lacked training and experience in video analysis, photography, and photogrammetry (she opined on all three of these disciplines in her report). She also didn’t follow best practices in the acquisition of the video, and didn’t apply any standard method to the photogrammetry. I provided a report that the prosecutor presented to the defense attorney, and the defense withdrew their experts report.

In a civil case that I previously posted about, the defense hired a photography expert to photograph the scene of an accident to illustrate the level of illumination. This expert photographed the wrong area, used the wrong equipment, and presented photographs that misrepresented the area. I was hired by plaintiff’s attorney to counter his testimony. I went to the scene and photographed the correct area following best practices and properly illustrated the illumination of the area. In addition, I provided cross examination questions, and the opposing expert basically admitted that he just followed directions of others and didn’t know if he photographed the correct area in the proper manner. This resulted in a multi-million dollar settlement for the plaintiff.

In a military court marshal case, the prosecution brought in an expert in photographic comparison. In this case, the expert didn’t follow standard protocols in their comparison and reached a conclusion that was well beyond what could be reached. The defense team brought me in to challenge this testimony. I provided the attorneys with the information they needed to understand the science of photographic comparison and to illustrate the weaknesses in the opposing expert’s report. I also provided cross examination questions, which were used, word-for-word, and resulted in the prosecutor asking defense counsel why they didn’t make a Daubert challenge. The reason was that they believed the strategy of showing the problems of the comparison was stronger then having it quashed in a Daubert hearing.

One last case was one in which the prosecution in a homicide case created a video to illustrate the illumination of the scene. A large segment of it totally misrepresented the scene, and the defense retained me to counter that evidence. I visited the scene and I wrote a report outlining all of the reasons why the police video misrepresented the scene. The defense presented my report to the judge and she ruled that the segment of video could not be used.

These cases involve three different aspects of my work – video analysis, forensic photography, and photographic comparison – but the common thread is that I was retained in each case because these attorneys needed someone to counter bad science presented by opposing counsel. In each case, the attorneys retaining me weren’t planning to use an expert until opposing counsel did so, and they needed someone to counter the bad science that was being presented.

“I need to tell others about you!”

I met with an investigator on a case last week to discuss her need for Forensic Video Analysis in a homicide case. During out meeting, she asked about the other work I do in photographic analysis and photography services. When I finished, she asked me for a handful of cards so she can let other investigators and attorneys know about my services – she said, “I need to tell others about you!” She had already viewed my website, but some of the things weren’t as clear as they could be, and I realized that I could do a better job in explaining what areas of expertise I work in, and when attorneys could use my services.

This post will just introduce this topic, and I’ll follow-up with more specific examples in the coming weeks. Here are the basics of what I do, in the coming posts I’ll discuss each area with examples from cases to illustrate how this applies to a variety of criminal and civil cases.

To begin with, my expertise is in forensic photography, photographic analysis, and video analysis.

In photography, I provide witness perspective photography, illustrative photography, accurate photographs of evidential items, injury photography, etc. I have photographed scenes to show evidence, to measure distances, to show what someone saw, etc.

In photographic analysis, I provide clarification of images that may be too dark, blurry, from a bad angle, etc. I also provide authentication – to determine if the image is a camera original; and if not, to determine if it represents what it purports to (if it’s been Photoshopped). And, I provide comparisons – is the subject in one photograph the same as a known object (same car, same person, same make and model of valve, etc.). I can also help determine the relationship of subjects in the photographs to each other.

In video analysis, I provide evaluation of the video (what is happening, and are there factors that can be causing someone to misinterpret the events in the video); comparison – is the subject in the video the same as the known object; clarification – stabilizing shaky video, deblurring, adjusting brightness and color, etc.); illustration – following the action in the video, synchronizing multiple video files together, combining video with maps, timestamps, or other events to give a more complete picture of the events.

And, of course, I can evaluate the work of opposing experts to see if they have provided a correct view of the evidence.

My next post will discuss a case I worked for the Los Angeles District Attorney’s Office recently that resulted in the defense withdrawing all of their proffered video evidence.

Expert Credentials

One of the first things that I get asked for when an attorney calls is a copy of my CV. An expert’s CV can tell you a lot, but when looking at a forensic photographer, photographic analyst, or video analyst, how should it be evaluated to make sure that you are getting a qualified expert? What should you be looking for on our CVs?

This week I’ll address work experience. In the coming weeks I’ll also discuss training, certifications, testimony experience, and some peripheral issues.

One thing to note is that there is a huge difference between the worlds of photography and video in the commercial world versus the forensics world. When searching for an expert, I would want someone with experience specifically in forensic or legal aspects of photography, or video analysis. Look at their work history and how many cases that they work in forensics. Ask how long they have worked in photography, photographic analysis, or video analysis; and what percentage of their work is forensic based. For me, I’ve been working in photography for over 30 years, and in forensics for over 20 years. I have fifteen years experience at a police department in the field of forensic photography, image analysis and video analysis.

Interestingly, the world of civil law and criminal law have their differences as well. I remember my first deposition in a civil case some eight years ago. It was unlike anything I had seen in the criminal cases I had worked. Your expert should understand their role in your case as well as the procedures regarding reports, depositions, court, etc.

The bottom line regarding experience – make sure your expert is experienced in their discipline, and also in the forensic application of that discipline. The workflow of the commercial photographer or post production video editor may be excellent in their own field, but may fall apart in a forensic environment.

Next week we’ll discuss what training to look for in your forensic video analyst, forensic photo analyst, or legal photographer.

In Photography and Video – We Want the Original Images

It always surprises me how often retaining attorneys send me digital copies of scanned color laser prints of photos in PDF format, or transcoded video files to do an analysis. Sometimes these images resemble their originals – sometimes they are a far cry from them.

To do the best enhancement, analysis, or authentication possible, originals are needed. Exact, logical copies of digital evidence is fine if the originals are digital.

With photographs, the original is the digital file as it came from the camera, or the negative from a film camera. If it’s a PDF, it is not the original.

With security digital video recorders, the original is an exact copy of the original data recorded by the DVR. If the file plays on a standard DVD player – it probably isn’t the original.

Determining if the still photos or video files are originals does involve more than simply determining whether the video plays on a standard DVD player, or if the still image file is in a JPEG file format – but those are good starts.

Whether you are using an expert or not – start demanding exact copies of the original visual evidence in your cases – it will make a significant difference in what that evidence can tell you.

Reflecting back on 2009

When I think about the past year, and cases I’ve worked, training I’ve received, training I’ve given, there are a few things that stand out.

In the arena of cases, I had two unique experiences – A Daubert Hearing in 12/08, and Grand Jury testimony in mid-2009.

The Daubert Hearing was on image authentication (are the images real, and do they represent real people and real events). That hearing was U.S. v. Frabizio in December of 2008, and was one of the few Daubert Hearings on photography, imaging, or image authentication. It was an interesting case, as it was going to its third trial. I was retained by defense counsel to analyze and interpret the government expert’s report and findings. I provided information to the defense attorney on this report, and attended the Daubert hearing, and answered her questions on his testimony. With the limited number of Frye and Daubert hearings on forensic photography, forensic image analysis, or image authentication, it was an interesting experience to participate in this hearing.

The Grand Jury testimony was also on an image authentication case. In this instance, I was an expert witness for the government. I have testified in cases since the early 1990s, but this was my first time to testify before a Grand Jury. It was quite interesting, as there is no judge, no defense attorney, and the accused is not present, nor is a case detective. It’s the prosecutor, a court reporter, the members of the grand jury, and the witness. In this case, I was impressed by the professionalism of the prosecutor and the members of the grand jury – I can imagine a situation in which a system without the formality or presence of a judge, defense attorney, etc. moving toward chaotic – I was pleased that this was far from the case.

As to training I received, it would be a toss up between two classes I took in the field of photogrammetry. One was on reverse projection photogrammetry that I posted about earlier. The other was a class in using Photomodeler (a software application for applying photogrammetry) for close range photogrammetry, presented by Dan Mills of DCM Technical Services. Photogrammetry is taking measurements from photos, and there are several approaches to this technology. Unfortunately, there are very few classes available in close range photogrammetry or in reverse projection photogrammetry. I was fortunate to get excellent training in each in 2009.

Regarding training that I provided – it’s difficult to isolate any one specific workshop. But, I did have the opportunity to give four workshops in digital photography to the San Francisco Police Department in 2009. That agency purchased Nikon D700 cameras for their crime scene investigators and wanted each member to be trained in the use of that camera for crime scene and evidence photography. I was proud to be the trainer they chose to provide this instruction. We held four, three-day sessions, with eight investigators in each session. It was a pleasure to work with this agency and to see their dedication to providing their front-line personnel with the tools and training to do their job.

Lastly, another highlight of 2009 was being appointed chair of the Forensic Photography and Imaging Certification Board of the International Association for Identification. It is an honor to have been chosen to fill this position, and our board is working hard on several projects.

The last year was successful in many ways – looking back at a few of the highlights helps see where we are today and prepare for the future. I’m looking forward to more interesting cases and great training opportunities in the coming year.


Reverse Projection Photogrammetry

Nov 22nd, 2009 - Filed under: Blog,Image Analysis,Photography,Video Analysis

The Law Enforcement and Emergency Services Video Association (LEVA) held it’s annual conference this week. I was fortunate to attend the two day pre-conference workshop on reverse projection photogrammetry presented by Richard Vorder Bruegge, Walter Bruehs, and Chris Iber, all from the FBI Forensic Audio, Video, and Imaging Analysis Unit.

Reverse Projection Photogrammetry is used to measure objects in a photograph or video. It can be used to determine if an individual could be excluded or included as a suspect based on height. In addition to using this method to determine the height of an individual in an image, I have used this technique to determine the position of objects in relation to their surroundings (after the objects were removed), the distance of safety barriers in construction scenes (after the construction was completed and the barriers removed), the size of cracks in a sidewalk (after the sidewalk was repaired), etc. from photos of those scenes.

The workshop covered issues related to the methodology and foundations of this technique. Any forensic image analyst or forensic video analyst who hasn’t yet had the opportunity to take this workshop should do so. It has been offered at the LEVA conference for two or three years, and also at the IAI Annual Conference.

So, What’s the Story on Resolution?

Nov 5th, 2009 - Filed under: Photography

My post on Synchronizing Video has generated a few comments related to image resolution. Three people, all of whom have a significant background in forensic imaging, have made rather divergent comments. Multiple topics are being addressed, from crime scene, accident and evidence photography to latent print photography to SWGFAST guidelines/requirements to the resolution of video security camera systems.

In all of these, there is a common element regarding resolution – the image should have the resolution to see the details for the purpose of the analysis. In a crime scene, this may be enough resolution to see the relationship of different objects in the image to each other. In a fingerprint, this may be second or third level detail. In the security camera it may be enough to read a license plate or see individualizing features in an object or person.

In security video, I almost always wish there was more detail, or higher resolution. These images are frequently 640 X 480 pixels or less. In addition, they are often highly compressed, the camera is at a bad angle, the lighting is poor, the optics are poor quality and dirty, and the subject of interest is frequently in the distance. Very few in the forensic video analysis field would argue that we’d all prefer security video with more resolution.

In fingerprints, I have seen some pretty good prints at resolutions as low as 300 PPI. The SWGFAST guidelines currently call for a minimum 1,000 PPI at a native resolution. So, can a fingerprint examiner use images that are less than 1,000 PPI or that may have been scanned with a scanner that interpolates when set to 1000 PPI? Of course they can – and they do.

Regarding crime scene, accident and evidence photography, many people compare digital imaging to film and suggest only using digital cameras that match the resolution of film. When I ask these proponents which film format (disc, 110, APS, 35mm, 6 X 6, 4 X 5), which film brand, which ISO, using which optics, whether they are concerned with tonal resolution as well as optical resolution, etc., they usually are unable to answer. There is nothing special about film that makes it a standard for measuring resolution.

When taking photographs, we simply need to know what level of detail is required, then be certain that we capture that level of detail. Arbitrary numbers, like the SWGFAST 1,000 PPI help us to guarantee that we are capturing enough detail, but we also need to understand that resolutions less than those recommendations don’t somehow become useless.