Abstract
Recently, the management of the database becomes more important. And it needs to find a method to search the queried image fast and correctly. Image database are usually implemented in two steps. First, we analyze the images in the database offline using some visual content attributes, and store the features to represent the image. Then, we apply the same analysis to the query image to compare the features to find out the similar or correct image. In the image retrieval method, there are many content attributes are adopted to measure the similarity between two images. In our method, we use the Sobel edge detection and proper threshold to detect the image to find out the image shape. Then, we use Hausdorff distance to compute the dissimilarity. We apply three measures to compute Hausdorff distance. First, we use gray level values to be the distance measure due to simple. Second, we use Euclidean distance. This measure can catch the exactly matching pixel location. The last one, we use the gradient of the image as the distance measure. This supports more information to be the comparison features; it can get the better query results.