Video

1. Abstract

This article outlines the technical principles, implementation methods.

2. Technical Principles

2.1 Optical Defogging

In nature, visible light is a combination of different wavelengths of light, ranging from 780 to 400 nm.

Figure 2.1 Spectrograms

The different wavelengths of light have different properties, and the longer the wavelength, the more penetrating it is. The longer the wavelength, the greater the penetrating power of the light wave. This is the physical principle applied by optical fog detection to achieve a clear image of the target object in a smoky or foggy environment.

2.2 Electronic Defogging

Electronic defogging, also known as digital defogging, is the secondary processing of an image by an algorithm that highlights certain object features of interest in the image and suppresses those of no interest, resulting in improved image quality and enhanced images.

3. Implementation Methods

3.1 Optical Defogging

3.1.1 Band Selection

Optical defogging is most commonly used in the near infrared band (NIR) to ensure penetration while balancing imaging performance.

3.1.2 Sensor Selection

As optical fogging utilises the NIR band, special attention needs to be paid to the sensitivity of the camera's NIR band in the selection of the camera sensor.

3.1.3 Filter Selection

Selecting the right filter to match the sensitivity characteristics of the sensor.

3.2 Electronic Defogging

The Electronic Defogging (Digital Defogging) algorithm is based on a physical fog formation model, which determines the concentration of fog by the degree of grey in a local area, thus recovering a clear, haze-free image. The use of algorithmic fogging preserves the original colour of the image and significantly improves the fogging effect on top of the optical fogging.

4. Performance Comparison

Most of the lenses used in video surveillance cameras are mostly short focal length lenses, which are mainly used for monitoring large scenes with wide viewing angles. As shown in the picture below (Taken from an approximate focal length of 10.5mm).

Figure 4.1 Wide View

However, when we zoom in to focus on a distant object (Approximately 7km away from the camera), the final output of the camera can often be affected by atmospheric moisture, or tiny particles such as dust. As shown in the picture below (Taken from an approximate focal length of 240mm). In the image we can see the temples and pagodas on the distant hills, but the hills below them look like a flat grey block. The overall feel of the image is very hazy, without the transparency of a wide view.

Figure 4.2 Defog OFF

When we turn on the electronic defog mode, we see a slight improvement in image clarity and transparency, compared to before the electronic defog mode was turned on. As shown in the picture below. Although the temples, pagodas and hills behind are still a little hazy, at least the hill in front feels restored to its normal appearance, including the high voltage electricity pylons further ahead.

Figure 4.3 Electronic Defog

When we turn on the optical fogging mode, the image style immediately changes dramatically. Although the image changes from colour to black and white (Since NIR has no colour, in practical engineering practice we can only use the amount of energy reflected by NIR to image), the clarity and translucency of the image is greatly improved and even the vegetation on the distant hills is shown in a much clearer and more three-dimensional way.

Figure 4.4 Optical Defog

Comparison of extreme scene performance.

The air is so full of water after rain that it is impossible to see through it to distant objects under normal conditions, even with the electronic defogging mode on. Only when optical fogging is switched on can temples and pagodas be seen in the distance (about 7km away from the camera).