Abstract: Residual film recovery is highly required for the high mechanical properties and picking-up rate with the low soil content of the recovered residual film. In this study, a roll-type picking component was designed to easily separate the soil. The rotary tillage blade was also used to throw up the membrane-soil agglomerate. The roll spring teeth were then selected to effectively pick up the residue film. The finite element method (FEM) and smoothed particle hydrodynamics (SPH) were coupled to establish the contact model between the roll spring teeth and the membrane-soil agglomerate. The finite element meshing was performed on the roll spring teeth and the residual film. The soil was assumed to the smooth particles. A microscopic analysis was made on the residual film stress and soil disturbance after the roll spring teeth were contacted on the membrane-soil agglomerate. It was found that the residual film stress was concentrated near the contact tip point of the roll spring teeth. The peak stress of residual film increased first and then decreased with the increase of the contact time. The maximum and average peak stresses of residual film were 0.2518 and 0.1319 MPa, respectively. The soil strain increased gradually with the increase of contact time, where the maximum soil strain was 3.271. The soil particles near the tip of the roll spring teeth were cracked to increase the soil disturbance. A three-factor five-level quadratic regression was conducted with the turning radius of roll, the diameter of spring teeth, and the top bend angle of spring teeth as the test factors, while the average peak stress of residual film and the maximum soil strain as the test indexes. A mathematical model was also established to clarify the influence of each influencing factor on the test indexes. At the same time, the optimal structure parameters of roll spring teeth were obtained as the roll turning radius was 100 mm, the spring teeth diameter was 5 mm, and the top bend angle of spring teeth was 42°. In this case, the average peak stress of residual film was 0.120 1 MPa, and the maximum soil strain was 3.758 4. The roll spring teeth were then fabricated using the optimal structural parameters after simulation. A new machine of residual film recovery was assembled to verify the influence of structure parameters of roll spring teeth on the picking-up performance. Field verification tests were conducted with the picking-up rate and soil content as the test indexes. A better performance was achieved in the picking up rate of 80.4% and soil content of 37.13%. There was no slip on the high-speed rotating of roll spring teeth. The residual film was constantly wound and indirectly punctured at the tip of the teeth. The roll spring teeth made it easier to crack the soil in the membrane-soil aggregate. The soil was then much easier to leave the residual film. The membrane-soil separation was gradually completed, indicating the soil no longer sticking to the residual film. Compared with the previous soil tank test, the picking-up rate increased by 8.74 percentage points, whereas, the soil content decreased by 12.18 percentage points, indicating the better performance of the machine. The finite element model can provide a strong reference for the structural optimization of the residual film recycling machine.