How Do You Solve the Problem of Bad Surveillance Video?By: John Honovich, Published on Jun 30, 2009
One of the most frequent complaints by both industry outsiders and megapixel camera vendors is how bad surveillance video is. It's so bad, many would say that they would challenge the basic point of even having such cameras.
Cases demonstrating this problem come up all the time. Here's an informative one recently from a California city:
It's almost impossible to make out who the criminal is in this video. To most people, this would be considered bad quality.
Is There Any Worth?
I still think there was significant value from this video for the authorities. They were able to pinpoint the exact time, the clothing of the suspect, gender and body type. All of these can be powerful tools in pursuing leads and narrowing down suspects.
What Do You Think?
What Can We Do to Improve?
First, getting good video quality in these conditions is not a simple manner of moving to megapixel. As we recently examined, the 'real' resolution of megapixel is routinely less than the pixel count stated and low light levels are problematic for megapixel.
There are some things we can do:
- The light level is almost certainly too low. To rectify this, we will have to add lighting or infrared illuminators. The problem with this is the expense. Minimally, costs will increase by $500 USD but depending on the range of illumination and the complexity of install, it could be significantly higher.
- The camera may need to be positioned differently to get a better view of the criminal. This is hard one to tell with the limited information I have. I suspect the camera was designed to capture vehicles coming down the alley. As such, the camera was not designed to handle this type of 'attack'. This, too, reflects a real world problem where cost constaints constrains the total number of cameras that can be deployed.
- The camera needs greater resolution - certainly. The question, though is how much resolution and how do we achieve it. The person's face looks to be 1/25th the width of the image or less. To see any details, we would likely need 300% (or more) increase in horizontal resolution. This is likely to push the limits of resolution levels we can achieve with security cameras today. Alternatively, we could narrow the field of view but we may need to give up the view of the street which the client may not want.