Traffic signs are one of the important roadway assets. Transportation agencies are required to inventory sign assets, but current manual traffic sign inventory methods are labor-intensive and time-consuming. A generalized traffic sign detection algorithm has been developed to automatically detect signs. However, correctly detecting sign images with discontinuous sign image boundaries (DSIBs) remains a challenge. This leads to false negatives (i.e., missing detection of signs), which critically affect the system reliability and hinder the implementation of an automatic sign detection system. This paper presents this critical issue of reducing false negatives. A hybrid active contour (HAC) algorithm is proposed with a new energy function on the basis of unique traffic sign characteristics, including location probability distribution function (PDF), statistical color model (SCM), and global curve length, to detect traffic signs with DSIB problems. The proposed HAC algorithm can be incorporated seamlessly into the existing sign detection algorithms to take advantage of the capability of the existing system while adding the strength of the HAC algorithm. The focused test shows that the proposed HAC algorithm can correctly detect 92% of sign images with DSIB problems that could not be detected previously. Using actual video-log images provided by two transportation agencies (607 and 1,547 images, respectively), the general test shows that the enhanced HAC-incorporated sign detection system can effectively reduce false negatives yet not add an excessive number of false positives. The false-negative rates decreased 6.8% and 9.2%, respectively, with minimal increase of false-positive rates. The preliminary results show that the proposed HAC algorithm has great promise for detecting sign images with DSIB problems.