Anfonwch hwn fel neges destun: Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding