Camouflaged object detection method based on homogeneous feature genetic algorithm

Camouflaged Object Detection (COD) is a challenging problem that affects both human and machine vision. Researchers have employed various feature enhancement techniques to capture hidden targets. However,these methods often overlook the disruptive effects of camouflage on detection performance. To address this issue,this study presents a novel method for camouflaged object detection based on homogeneous feature genetics,aimed at removing camouflage and enabling multi-view target detection. In the proposed method,a Transformer architecture is used to extract multi-level features,and target manifold skeletons are generated through wavelet transforms and heterogeneous gradients. Finally,a multi-view genetic fusion strategy is designed to achieve camouflaged target focusing and prediction. Experimental results demonstrate that the proposed method exhibits strong generalization and robustness and provides clearer segmentation for camouflaged object detection in complex backgrounds.