Alleviating Hallucinations in Large Vision-Language Models through Hallucination-Induced Optimization