In this paper, we examine the suitability of correlogram for background subtraction, as a step towards moving object detection. Correlogram captures inter-pixel relationships in a region and is seen to be effective for modelling dynamic backgrounds. We propose herein a novel feature, termed fuzzy correlogram, composed by applying fuzzy c-means algorithm on correlogram. Fuzzy correlogram greatly reduces the number of correlogram bins and the quantization noise. The approach handles multi modal distributions without using multiple model features, unlike traditional approaches. Effectiveness of the proposed method is illustrated on different video sequences.