ページ1に含まれる内容の要旨
Tech Note
TN-0902 Date: 02/06/09
UsiNg a 2CCD Camera To Crea Te HigH-DyNamiC raNge images
Some imaging scenarios push dynamic range beyond the capabilities of the typical sensor. This is especially
true where incident light is present (e.g., imaging a light source and the surrounding area). This can also
occur in situations with bright reflections or in high contrast indoor/outdoor scenes where one needs to cap-
ture details in both bright sunlight and dark shadows. One technique for
ページ2に含まれる内容の要旨
Tech Note Example 1 - Maximum Dynamic Range (no overlap) To create an image that spans the maximum range of light intensities, use shutter settings to calibrate the 2 sensors so that Sensor B = Sensor A * 1024. In other words, the light needed to generate 100 counts from Sen- sor B is 1024 times the light needed to create 100 counts from Sensor A. For example, if Sensor A is operating with the shutter off (1/30 sec.), Sensor B would need to be set as close as possible to 1/30720 sec. using th
ページ3に含まれる内容の要旨
Tech Note Displaying a high dynamic range image on a standard monitor will require mapping the output to fit the monitor’s dynamic range capability. For the image described above, start by creating a 20-bit image map using the raw pixel data. Then create a 10-bit image map to display on the monitor by dividing the 20-bit im- age map by 1024. Or create a 12-bit image map by dividing the 20-bit image map by 512. The preceding process describes using both output channels in 10-bit mode. Alternati
ページ4に含まれる内容の要旨
Tech Note Now, our post processing routine could be handled as follows: if (pixel B < 16){ pixel_out = pixel A }else{ pixel_out = pixel B * 64 } By overlapping the two sensor responses, this approach utilizes the full precision of the lower 10-bits while reducing the effect of noise at the transition point and greatly increasing the precision of the upper 6-bits. Example 3 – Averaging the Overlap Both of the preceding examples assume that a precise calibration can be made between
ページ5に含まれる内容の要旨
Tech Note Now as Sensor A approaches its saturation point (512 – 1023 counts) the output uses the average of both sen- sors’ data to “smooth” the transition between the two sensor response graphs (see Figure 6). It still limits the use of the lowest bits on Sensor B (those that are most susceptible to noise) and keeps the calibration factor at 64 to increase the output precision of the upper bits. FIGURE 6 – Averaging used to smooth calibration in overlapped region Example 4 – Dual-Slope Dyna
ページ6に含まれる内容の要旨
Tech Note For example, in a 4-bit overlap scenario, we still set the shutter for Sensor B to be roughly 64 times faster than Sensor A (i.e., a pixel on Sensor A with a value of 256 would have a value of 4 on Sensor B). But, de- pending on our objective, we apply a post processing factor of less than 64 and add an offset value to get the two output graphs to intersect somewhere around 768 counts. Finally, use a post processing algorithm that averages the values of Sensor A and B in the last bit
