This paper presents an automatic sea surface object clustering and tracking in forward looking sonar images. The considered sea surface obstacles are man-made objects: buoys, boats, ships (motorboats or sailboats). Their acoustic signature varies according to their type and state (fixed or moving). The proposed method detects the various target signatures. Then detections are gathered into clusters using an automatic and adaptive clustering method based on the Delaunay Triangulation. After the clustering stage, isolated detections are rejected and the remaining clusters are classified into two types: clusters including wake or not. This classification is based on the shape eccentricity feature. Then, according to the cluster type, the obstacle position is extracted to be used for tracking. For clusters without wake, the obstacle position is the centroid of the cluster but for clusters including wake, wake extremities are set as possible vessel position. Finally, obstacle tracking is carried out in Cartesian coordinates using the Debiased Converted Measurement Kalman filter and the Joint Probabilistic Data Association Filter. Promising results are obtained using real data collected at sea with various objects and scenarios.