to the work aims to improve the correct recognition rate and robustness of hash sequences for tampered content by proposing a robust hash algorithm based on color vector angle histogram and coupled discrete cosine transform compression. Combining interpolation operation with Gauss filter, the image was preprocessed to produce a fixed-length hash sequence for any authentication target. The polar coordinate transformation LPT (Log-polar transform) was introduced to transform the coordinates the image in standard size to output the secondary images. SVD method (Singular Value Decomposition) was used to process the secondary image to output global robust features. Subsequently, the color vector angles of all the pixel components in the normalized size image were extracted, and the corresponding histograms were calculated to form color perception features. Combined global and color perception features were regarded as 1D hash sequences. Based on discrete cosine transform (DCT Discrete Cosine Transform), the 1D hash sequence was compressed to get a set of AC coefficients. By introducing the logistic map, the differential encryption method was designed by the chaotic sequence to diffuse the AC coefficients to form the target hash sequence. Based on the optimal decision threshold and l2 norm distance, the authenticity of detected image content was dis-criminated. From the test results, the proposed scheme had stronger robustness and higher detection accuracy for various geometric attacks than the current hash mechanism. This hash method has high robustness and recognition accuracy, which has good reference value for trademark retrieval, information anti-counterfeiting and other fields.