关键词: 聚类分析, 图像分割, 文字提取

Abstract: To deal with the gradient problem in the clustering process of text extraction, an algorithm based on binary clustering was proposed. The original image was converted to binary bitmap after preprocessing. The background blocks of the image were clustered by the region features, and then text blocks were recognized by the distribution features. The experiment shows this method achieves satisfactory result on various kinds of images.

Key words: clustering analysis, image segments, text extraction