Most　of　the　current　image　watermarking　algorithms　focus　on　robustness　and　imperceptibility,　while　few　research　targeting　security.　This　paper　combines　digital　watermarking　with　cryptography,　and　proposes　an　image　encryption　algorithm　combined　Memristive　Chaotic　Cellular　Neural　Networks　(MCCNN)　and　Arnold　transformation.　The　MCCNN　is　used　to　generate　multiple　and　diverse　chaotic　sequences　to　improve　the　security　of　the　watermark　image,　and　the　two-dimensional　discrete　wavelet　transform　(DWT)　is　adopted　to　embed　the　watermark　according　to　the　embedding　intensity　value.　The　experiment　results　show　the　effectiveness　of　the　proposed　image　watermarking　algorithm　based　on　double　encryption,　which　is　robust　to　common　attacks　such　as　image　rotation,　noise　addition,　and　JPEG　compression.　©　2023　IEEE.