In this paper, a method of self-recovery and tamper detection applied to color images is proposed. Designed framework divides the image into non-overlapped blocks, and for each block watermarks are generated, which are applied to tamper detection and recovery. In order to avoid the tampering coincidence problem, generated watermarks by each block are embedded using the LSB method in a different block in an embedding sequence, which is generated by a seed. During the authentication and recovery process, previously embedded watermarks in each block are extracted and can locate modified regions, following recovering the manipulated block via usage its recovery watermark. To obtain a better performance the designed scheme embeds three versions of the watermarks generated for each block. In simulation, the quality metrics PSNR (Peak Signal-to-Noise Ratio) and SSIM (Structural Similarity Image Measure) were used, justifying better performance during the recovery process at different tampering rates via comparing the designed scheme with similar state-of-the-art methods.