There are different aspects for classification of SVD-Based watermarking schemes, which can be seen in The Fig. In accordance with the domain, there are four types of domain that SVD is applied on them, such as Spatial Domain, discreet wavelet transform (DWT), discreet cousin transform (DCT), Hybird-DWT&DCT , and invariant wavelet transform domain (IDWT). In spatial domain techniques, SVD decomposition is applied directly into the cover image’s pixel values, and the watermark is embedded by modifying the coefficients of SVD decomposition of the cover image. In Frequency Domain techniques, the cover image is fragmented into a variety of frequency bands. Different techniques used diverse reversible transforms, such as Discrete Cosine Transform (DCT), or Discrete Wavelet Transform (DWT), Hybrid-DWT & DCT and invariant wavelet transform domain (IDWT). Here, SVD is applied on its transformed representation bands of the cover image and then the watermark signal is inserted by means of modifying the coefficients of SVD components. From AI approaches point of view, there are several techniques successfully implemented AI population-based algorithms on SVD based image watermarking schemes. In such techniques; automatically the suitable scaling factor, threshold, compensation, are chosen for modifying the coefficients of SVD components. Such as Genetic algorithm (GA), Differential evolution (DE), Firefly algorithm (FA), K-means, Artificial Bee colony (ABC) and Particle swarm optimization (PSO).Based on the modifying component, as shown in Fig. 3, some SVD- based watermarking schemes modify S component (singular values) for inserting watermark signal. These techniques try to embed an adoptable watermark image that has the same range as S coefficient values, thus the damage is less due to the modifying. Some other techniques use U component or V T component and both U & V components [6, 18, 29] to embed watermark. In such strategies a row or column of SVD coefficient components are chosen to be modified. Finally, based on the last classification, SVD coefficients can be directly modified by adding the watermark values to them, the modification is based on a scaling factor to trade off between the quality and robustness of embedding. Some other schemes employ indirectly coefficient modifying. Instead of adding the watermark to coefficient values , the relationships between the coefficient vectors are changed for inserting the watermark signal based on a specific equation