Shared images obtained from the Web and social platforms could be degraded by several operations, compromising their quality and reliability for consumers. Existing image restoration algorithms primarily focus on single degraded images with known degradation types, while limited research has been done on multi-degraded image restoration with unknown degradation types. Additionally, existing methods assess image reliability by detecting whether the image has undergone specific single operations using classification models. However, accurately estimating the complete operation history of an image still remains challenging. These limitations may restrict their practical use in consumer applications. To address these issues, in this work, we propose a novel framework that utilizes operation history estimation to enhance the reliability and quality of multi-degraded images. First, we develop a deep network based on the machine-translation mechanism to estimate the operation history experienced by the input multi-degraded image, including operation types, parameters, and their order. Specifically, a self-attention scheme and a cross-attention scheme are designed to capture both the internal correlations among operations and parameters and their interactive relationships. Second, we propose applying our operation history estimation scheme to multi-degraded image restoration, utilizing the estimated operations and associated parameters as prior information to assist the restoration process. Extensive experimental results demonstrate that our framework achieves excellent performance in operation history estimation and multi-degraded image restoration, ultimately improving the reliability and quality of images in consumer applications.