Feedback Structures as a Key Requirement for Robustness: Case Studies in Image Processing

Natural as well as technical systems owe their robustness to a large extent to feedback structures. To use feedback, it is necessary to define a measurement, an actuator in the system, and a setpoint or reference—the measurement is compared with the reference and an actuator action is derived from the difference between them. This simple but powerful structure is responsible for providing the system with robustness against external influences. For a detailed discussion, the image processing system is chosen. The robustness of an image processing system is considered here to be the ability of an algorithm to achieve the desired output independently of numerous external influences such as illumination conditions, the imaging system, and imaged objects characteristics. Because of a number of problems, such as the absence of feedback from the higher to the lower processing levels, a traditional image processing system is of low robustness. This chapter presents the novel idea of the inclusion of feedback control at different processing levels to overcome the above problems of traditional image processing. The main idea behind this is to change the processing parameters in a closed-loop manner so that the current processing result at a particular processing level is driven to a reference result, providing the subsequent higher processing level with reliable input data. Presenting image processing as a new control application field, the chapter focuses on the specific features of image processing that make closed-loop control in this area different from conventional industrial control. The advantage of feedback for this advanced, prominent, and important application area is demonstrated through two examples.