A distance transform of binary images is an operation which computes the shortest distance between a background pixel and a foreground pixel. Distance information from this transform is different according to distance metrics. There are many distance metrics, such as Manhattan distance transform(MDT), Chessboard distance transform(CDT), Euclidean distance transform(EDT), Squared Euclidean distance transform(SEDT), and so on. Especially, SEDT is commonly used due to efficiency of calculation and fast executive time. Methods to obtain the distance transform are mostly classified into four: parallel, sequential, propagation, and linear time separable methods. Linear time separable method(Meijster’s) is virtually the best performing method of them. Linear time separable method consists of two phases. The first phase scans an image row-wise, while the second phase scans the image column-wise. SED is computed by finding the lower envelop parabola(LEP) at the second phase scans. LEP is determined by comparing heights of two parabolas and performing an intersection operation in order to limit the range of the lower parabola. In this paper, a method improving the column-wise operation of Meijster’s algorithm is proposed. An intersection point, which is located between consecutive background pixels, is always in the middle of the two parabolas and then, a starting point of LEP corresponds to the middle point in the case of continuous background pixels. The proposed method determines LEP without the height comparison between parabolas and the intersection operation in consecutive background pixels. Thus, the proposed method can reduce these intersection operation times in continuous background pixel. In order to evaluate the performance, the proposed method is compared with two conventional methods through six images. As a result, it has shown that the proposed method decreased greatly the performance time compared with the others.