Abstract: In the rice-oil rotation regions of the middle and lower reaches of the Yangtze River, the prevalence of heavy, sticky clay soil presents significant challenges for mechanized rapeseed planting. Conventional profiling mechanisms often struggle to maintain consistent bed shaping, leading to severe soil adhesion, gully formation, and clogging of the machinery. These issues result in uneven seedbed surfaces, water accumulation, and waterlogging stress, which in turn compromise the stability of seeding depth and the overall quality of rapeseed establishment. To address these agronomic and mechanical bottlenecks, this study investigates the anti-adhesion mechanism of a rotary compressing and shaping device designed for overall bed profiling. The research begins with a mechanical-soil interaction analysis of the rotary compressing process. Theoretical results indicate that during the operation of a passive (non-driven) rotary roller, the contact area between the roller surface and the soil decreases while contact pressure increases, leading to inevitable soil accumulation and difficult detachment. In contrast, an active (driven) rotary roller generates a relative velocity difference between its surface and the seedbed. For forward rotation, the already-formed bed exerts a shearing and squeezing action on the roller surface, creating an active anti-adhesion mechanism. Conversely, while a reverse rotation roller utilizes the unformed soil to scour the roller surface—effectively preventing adhesion—the opposing motion and excessive grinding force tend to induce defects such as surface cracks on the seedbed. To visualize these microscopic interactions, a coupled DEM-MBD simulation was conducted. The simulation analyzed the velocity, displacement, and kinetic energy changes of soil particles adhering to the active roller. It revealed that soil adhesion primarily initiates in the contact zone with the unformed, loose soil, whereas the active anti-adhesion effect predominantly occurs in the contact zone with the compacted, formed bed. Subsequent bench tests corroborated the theoretical and simulation findings. Results showed that for the forward-rotating roller, soil adhesion area, adhesion volume, and bed surface roughness followed a trend of initially increasing and then decreasing as rotational speed increased. Notably, the anti-adhesion performance was poorest when the speed ratio was near 100% (where the linear velocity matches the forward speed). For the reverse-rotating roller, while adhesion decreased with higher speeds, the bed surface roughness increased significantly due to soil disturbance. The passive roller demonstrated the worst performance in both anti-adhesion and shaping quality. Further optimization was conducted using orthogonal bench tests, with roller speed and absolute soil moisture content as experimental factors. The evaluation indices included adhesion area, adhesion volume, and surface roughness. The results indicated that both rotational speed and soil moisture significantly affect adhesion performance. Crucially, the forward-rotating roller demonstrated superior adaptability to variations in soil moisture content compared to other configurations. Finally, field trials were conducted in post-rice stubble fields with sticky soil (absolute water content >30%). The forward active rotary roller achieved a soil adhesion area of 62.4 cm² and an adhesion mass of 174 g, representing a reduction in adhesion of 90.05% and 87.77%, respectively, compared to the passive roller. The bed surface roughness was recorded at 15.7 mm, an improvement of 86.08% over the passive counterpart. In contrast, the reverse-rotating roller resulted in excessive grinding of the bed surface. This study confirms that the forward-rotating active profiling mechanism is the optimal solution for sticky soils, providing a critical reference for the design and improvement of seedbed preparation equipment in the Yangtze River basin