For safe navigation of visually impaired, an assistive system that can estimate the distance betweenvisually impaired and obstacle and can intimate the user is needed. Estimation of the distance betweenan obstacle and visually impaired (user) is challenging due to artifacts in the real environment such asvariation in speed of obstacle and non-uniform illumination conditions. This paper presents a novelalgorithm to estimate the distance of an obstacle from the user using Speeded Up Robust Features(SURF). Instead of traditional distance measurement sensors, SURF features are used for distancemeasurement of an obstacle from visually impaired. The input video frames are preprocessed, andcorrection for non-uniform illumination is applied. The dominant points in each input video frame arelocated. The correspondence between the dominant points in successive frames is derived. For a typicalcamera, the average magnitude of SURF features is a linear function of a distance between the user andthe obstacle. This function is used to estimate the distance between obstacle and user. The proposedalgorithm is tested on videos recorded in a dynamic environment. For videos captured with Microsoftwebcam, an average % error for distance estimation is 1.31%, and for speed, estimation is 4.18%. Forvideos captured with an Iball camera, the average % error for distance estimation is 2.134%, and forspeed, estimation is 0.399%. The performance of the proposed algorithm is compared with existingtechniques on the basis of an error in distance estimation, standard deviation, space complexity, andtime complexity.