The Causal Real-time Relationship Reed-X Detector (CR-R-RXD) detecting algorithm based on the pixel-by-pixel processing for hyperspectral imagery, which has the problems of a large amount of computation, a long display time and a slow running speed. A CR-R-RXD detecting algorithm based on line by line was proposed in this paper. Compared with the CR-R-RXD method based on pixel by pixel processing, the whole row pixel vector of hyperspectral image was used as input in this proposed method. That is, dealing with a row of hyperspectral data needs to be calculated only once, which greatly reduces the calculation times. Experimental results show that, to compare with the R-RXD algorithm and CR-R-RXD method based on pixel by pixel processing, the proposed algorithm can achieve the process of fast real-time processing with almost the same accuracy as the R-RXD algorithm, the detection accuracy is improved to compare with the CR-R-RXD algorithm based on pixel by pixel processing, and the testing time of the algorithm is reduced, which enhances the timeliness of the algorithm.