The concepts of uncertain pixel and pattern are used to improve the distortion-invariant ability of morphological hit-or-miss transformation (HMT). A method called joint rank-order multi-value hit-or-miss transformation (JRMVHMT) is introduced, which reduces the decision effect of the pixels that is easily disturbed by rotation and scaling distortion. Because of not losing information of the input images, JRMVHMT can realize the precision recognition between two images and has a better distortion-invariant ability for scaling and rotation than the ordinary HMT. The comparison of JRMVHMT with the ordinary HMT in theoretic analysis and experiments is given. Based on an incoherent correlator and multil-value complementary encoding, JRMVHMT is realized for binary image recognition and gets satisfying results.