What is fine resolution?

The technology used to enhance the resolution of a digital image is known as super resolution. Over the course of digital imaging system development, many different techniques of super resolution have been developed; each with its advantages and disadvantages. A thermal imag­ing system assimilates the resolution enhancement technology used in digital imaging system to improve on its resolution. The TrueIR thermal imager uses a specific multi-frame super resolution technique and algorithms that is known as fine resolution, which enhances the resolution of a thermal image by four times.

So what is fine resolution and how does it work?

It can be summarized to 3 main processes – Acquisition, Super-position, Reconstruction.

Each of the above mentioned processes have different tasks on its own.

Acquisition:

Whenever the trigger button is pressed, the thermal imager will automatically capture multiple images of the same scene continuously. It also assumes that each of the images or frames taken is shifted, due to natural hand movement of the thermographer. Also, during the acquisition, all the frames are automatically up-scaled to higher resolution images through interpolation technique – predicting image pixels using data from adjacent pixels that has been captured initially. The technique is very similar to a curve fitting mathematic function. The new pixels are predictive values instead of measured values.

The interpolation process is fast, hence, the high resolution inter­polated images are real time. Instead of showing the real time low resolution image on the display, the interpolated high resolution images are displayed on the LCD, which serves as a view finder for the thermal image, just like a digital camera.

Super-Position:

As mentioned, each of the frames taken through multi-frame acquisition process is slightly shifted. Therefore, overlapping the frames without any intelligence will not work. In this process, feature points from each of the frames needs to be identified, positioned and aligned together before being superimposed to form a high resolution image. Figure 1 below illustrates this process.

Figure 1: Super-position

Reconstruction:

Superposition process above might incur some noise into the thermal image, causing it to be fuzzy. Hence, at this stage, many mathematical models and image processing techniques are used, such as an averaging algorithm for noise reductions and an edge enhancement algorithm for image sharpening.

The Result: 

Here’s a simple lab test to prove the point. We simply measure the temperature of a slim 1-mm vertival bar at a fixed distance. Figure 2 illustrates the ability of the TrueIR detector’s array to capture the thermal image of the slim bar. Due to limitation of the physical iFOV discussed earlier, if a single frame is used (in this example that means only Frame 1 is used), only the average temperature of the bar is recorded, which is inaccurate.

Figure 2: Detector's pixel arrays

Through multi-frame acquisitions, Fine Resolution is able to recover sub-pixel information. Figure 3 shows the test result comparison between using a detector with 160 x 120 pixels versus the results obtained using a fine resolution imager which generates an image with 320 x 240 pixels. Fine resolution provides 1.5x better iFOV, thus resulting in 1.5x more accurate temperature measurement. So you get to reap the benefit of a 320 x 240 pixels thermal imager for just a fraction of the cost!

Figure 3: Lab test resuts (1-mm bar)

Sample Fine Resolution IR images taken with Keysight's U5855A TrueIR thermal imager

To read more about Fine Resolution capability, click here. 

The U5855A TrueIR thermal imager

Or to learn more about U5855A TrueIR thermal imager, go to Keysight U5855A TrueIR thermal imager