SENSORS
How Image Signal Processor Tuning (ISP Tuning) works and what it is good for
Wherever images are taken, image raw data is generated. This article explains why this raw data is post-processed by internal algorithms in cameras and why manual adjustment is also necessary for industrial applications. First of all, it is important to understand how a sensor is structured and what an image signal processor actually is.
T
his is how (CMOS) sensors are constructed. A standard CMOS sensor consists of photocells that absorb light; the pixels. A hypothetical image sensor with 64 megapixels could, for example, consist of a matrix of 8000 x 8000 pixels. However, each pixel can only perceive brightness information. To enable colour images to be or blue, is placed over each pixel. The colour calculation of the Image Signal Processor (ISP) begins.
The Image Signal Processor The ISP is basically nothing more than a microchip that performs tasks like a CPU in a laptop. In contrast to the CPU, however, the ISP specialises in imaging calculations. The ISP uses the brightness information of the pixels to calculate an entire image. However, if only this information were available, the result would be a rather dull monochrome image. The algorithm that the ISP uses to create a colour image is called the debayering algorithm, better known as demosaicing. This generates a colour image. The sensitivity to light differs from sensor to sensor. To account for this and to obtain a colour photo with usable colours, another algorithm is now applied; the colour-correction matrix algorithm. There are many other algorithms that the ISP executes to improve image quality. In addition to the colour correction and demosaicing mentioned above, these include automatic white balance, automatic exposure, gamma correction, HDR, lens shading correction, lens distortion correction, defective pixel correction and noise reduction.
26 APRIL 2025 | ELECTRONICS FOR ENGINEERS
Why tune the ISP?
While the quality of an image is perceived subjectively, different standards from ISO, CIE different applications. For example, ISO 12232 for sensitivity, ISO 12233 for resolution, ISO 14524 for photon conversion characteristics and CIE for color reproduction quality. The or more of the standards. However, ISP For example, one could enhance the red for a medical camera to make the bleeding more evident to the surgeon.
This is how the ISP is tuned Since every sensor is different and every application has its own special requirements and environmental conditions, there are imaging laboratories that use a standardised approach to determine the then adjust the ISP parameters accordingly. These labs examine the results of the ISP algorithms with default parameters on the basis of the sensor characteristics, the selected lens and the given application by means of tests. The results are used
to calculate the needed correction to be applied to the parameters. This is the basis for tuning.
For example, vignetting or unevenly distributed brightness in an image, is removed; by testing different exposure scenarios, the correct values are determined to create evenly distributed image brightness in a particular situation. Various lighting scenarios such as D65 (average daylight, colour temperature colour temperature 4000K) are also tested for white balance, and the appropriate ISP parameters are found for the application. This iterative process gradually results in a complete set of ISP parameters that is
Correct ISP tunes are crucial to the success of such models. If the environmental or application conditions are well known, the right settings can be made in advance and the software can be adapted accordingly. Important parameters include black level correction, lens shading correction, white balance, noise reduction and colour correction. Extensive testing is required to application. This includes environmental conditions and software.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50
