Profiling headers, agricultural machinery capable of automatically adapting to ground undulations, play a significant role in the recovery of cotton stalks. This study investigates the application of profiling headers in the cotton stalk recovery process, designs the structure and hydraulic system of the profiling system, and proposes two control strategies: PID control and fuzzy control. Comparative analysis reveals that fuzzy control outperforms in terms of stability, accuracy, and responsiveness, making it suitable for use in farmland environments with significant nonlinear disturbances. This paper also explores methods for measuring cotton stalk height, utilizing the Yolov5s object detection model combined with the SGBM stereo matching algorithm to achieve accurate measurement of cotton stalk height. Experimental results indicate that fuzzy control enables the header to intelligently respond to ground undulations, with system response times meeting the requirements for cotton stalk recovery. Additionally, the height measurement using a stereo camera reduces manual measurement time and provides valuable insights for header blade speed adjustment and blockage prevention. This research is of great significance for the intelligent development of cotton stalk recovery machine headers.