圖像加密是防止圖像中個人隱私洩露的解決方案，但由於加密後的圖像沒有意義因此很難獲得準確的圖像，為了找到圖片，我們會將索引值嵌入到加密圖像中，未來要查找這張圖片的時候，可以透過取出隱藏在圖片中隱藏的數值就能知道哪些圖片是我們要查找的，因為影像的恢復是非常重要的，所以我們的方法是在取出隱藏數值之後可以通過預測和結合漢明碼或是直方圖來恢復圖像。我們提出了兩個方法將祕密訊息嵌入到加密後的影像裡，第一個方法是透過漢明碼結合預測的方式，但是在預測的時候會有兩種結果導致錯誤的產生，為了使解密後發生預測錯誤的機率降低，所以我們會使用Location Map的方式，如此一來就能降低錯誤的發生與達到恢復圖像的作用。第二個方法則是使用了直方圖的方式，直方圖的優點在於擁有很低的運算量，可以讓我們快速地且有效率的使圖像在解密後能無損的恢復成原始的圖片。
Image encryption is a solution to prevent personal privacy leaking in images, but since the encrypted image is meaningless, it is difficult to obtain an accurate copy. To find the search target images, we embed the index values into the encrypted images. To get the search targets, you can extract the hidden data from the images to make sure whether they are the exact ones as your wishes. Also, reversibility is an ability to restore the embedded images into the original ones, which are essential to the specific domains, such as military, medication, and so on. In the thesis, we propose two methods to encrypt the images as well as to embed the secret messages into the encrypted images. The first method is to combine the prediction method with hamming code; the prediction errors are kept in a location map for later wholly restoring. The second method encrypts and embeds by skillfully using the characteristic of the histogram. The advantage of the second method is that it is with a low computation load, which allows users to quickly and efficiently restore the image to the original image without any losses after decryption.
