国家自然科学基金资助项目（51208130）；西部交通建设科技项目（2007 318 223 01 9）
To overcome the shortcomings of traditional methods for joint sealant rupture evaluation, a new method for recognition joint sealant rupture of cement concrete pavement based on digital image was proposed. Based on the digital image processing in accordance with the joint feature,the slab joints can be positioned. The joint sealant rupture can be classified by pattern recognition according to the feature of images of joint sealant rupture. Joint positioning was divided into two main steps：approximate location based on the gray projection feature of whole image and accurate location based on the character of gray projection and edge projection around the joints image. In the process of pattern recognition, the features including the relative width of joint sealant interspace, the relative location of joint sealant interspace, the relative width of outer rectangle of joint sealant interspace and the relative projection value of joint sealant interspace were extracted from the digital images. BP network and linear discriminant were adopted for the classification.The experimental results show that accuracy of recognition on joint sealant rupture of cement concrete pavement reaches 94%.