Research on the Sharpness Evaluation Function Based on GrabCut for Autonomous Focusing in the Fluid Physics Experimental Cabinet of the Tiangong Space Station

Autonomous focusing in aerospace systems often relies on image sharpness evaluation algorithms, which are primarily optimized for satellite cameras and thus unsuitable for experimental images from space stations. To address the lack of autonomous focusing capability in the fluid physics experimental cabinet aboard the China Space Station, this paper presents a novel image sharpness evaluation method based on an edge gradient function utilizing GrabCut. The proposed method is specifically tailored to the unique characteristics of images from the experimental cabinet, where color blocks are concentrated, the observation area is central, and boundaries are distinct in clear images. These distinct features enable the method to effectively leverage edge information for precise evaluation, making it particularly suited to the requirements of the Space Station’s experimental environment. We employ the GrabCut algorithm to segment the foreground from the background, using the boundary as the target area for gradient computation. A feature-based generic initial bounding box is adopted, ensuring broad applicability to all images from the fluid experimental cabinet without requiring user intervention. Subsequently, a convolution operator calculates the gradient along the target area, which serves as the sharpness evaluation function. Compared with two commonly used algorithms, experimental results show that the proposed algorithm achieves clear unimodality near the focal point and remains robust against blurring and noise in out-of-focus regions. Consequently, this algorithm has meaningful engineering potential, satisfying essential sharpness evaluation requirements and providing an effective solution for image sharpness assessment in the Fluid Physics Experimental Cabinet of the Tiangong Space Station.