Enhancing Visual Document Understanding with Contrastive Learning in Large Visual-Language Models | IEEE Conference Publication | IEEE Xplore