The major issues in most watermarking schemes are security, reliability, and robustness against attacks. To achieve these objectives in a watermarking algorithm, the selection of a scale factor to embed the watermark into the host image is a challenging problem. In this paper, a block singular value decomposition (SVD)-based reliable, robust, secure, and fast watermarking scheme is proposed that uses particle swarm optimization (PSO) in the selection of the scale factor. SVD is applied here on the nonoverlapping blocks of LL wavelet subbands. Selected singular values of these blocks are modified with the pixel values of the watermark image. Selected locations of these singular values increase security. In addition, the scale factor using PSO for embedding the watermark increases the robustness and imperceptibility of the proposed scheme. Direct embedding of the watermark image into the host image and the use of block-SVD makes the scheme faster. Comparative analysis with an existing algorithm shows that the proposed technique performs well during most types of noise attacks and removes the diagonal line problem present in the extracted watermark image.