Multimodal biometrics systems ensure high-level security for critical real-time applications in comparison to unimodal systems. However, adaptive integration of multi-modality and its resistance to various attacks is a foremost concern. In response to these challenges, an adaptive multi-biometric system is proposed that integrates multiple modalities namely, finger and iris dynamically. The feature scores for each modality are computed and integrated adaptively using the proposed fusion approach. To compute optimal modality scores, scores are optimized using crow-based optimization. The proposed method is evaluated across multiple datasets to showcase its effectiveness and efficiency, with the results averaged for a comprehensive assessment. It has achieved superior performance compared to other state-of-the-arts approaches in terms of accuracy and equal error rate (EER).