The key techniques on ROI confirmed and similarity computation in ROIBIR system are studied. Firstly, two approaches of confirming ROI are introduced by considering the visibility of image regions, the possibility of designating region and its weight for users making the image retrieval system include human perception and accord with human vision habits. Then, a region-based similarity measurement approach is introduced. With the proposed approach, the similarity of two regions with the combination features, histogram and region shape features is computed respectively. Then, the similarity of two regions is computed by the weight product method. The average value's similarity for every region is regarded as the similarity of ROI and target image. Experiments show that the proposed methods are available.