A Computational Framework for Apple Detection Using Fuzzy Logic and Structural Cues

Accurate and reliable detection of apples in complex orchard environments remains a challenging task due to varying illumination, cluttering backgrounds, and overlapping fruits.In this paper, the difficulties were tackled with a novel edge-enhanced detection framework proposed to integrate dynamic image smoothing, entropy-based edge amplification, and directional energy-driven contour extraction.An adaptive smoothing filter was adopted with a sigmoid-based weighting function to selectively preserve edge structures while suppressing noise in homogeneous regions.The input of Red Green Blue (RGB) image was subsequently transformed into the Hue, Saturation, and Value (HSV) color space to exploit hue information, thereby improving color-based feature discrimination.The introduction of a hybrid entropy-weighted gradient scheme helped strengthen edge detection, that is, the local image entropy modulated gradient magnitudes to emphasize structured regions.A global threshold was then applied to refine the enhanced edge map.Ultimately, continuous apple contours were extracted using a direction-constrained energy propagation approach, in which connected edge pixels were traced according to compass orientations, thus ensuring accurate contour assembly even under occlusion or low contrast.Experimental evaluations confirmed that the proposed framework substantially improved the accuracy of boundary detection across diverse imaging conditions; its potential application in automated fruit detection and precision harvesting was therefore highlighted.