In this paper a DWT based steganography in frequency domain, termed as ATFDWT has been proposed. Here, the cover image is transformed into the time domain signal through DWT, resulting four sub-image components as 'Low resolution', 'Horizontal orientation', 'Vertical orientation' and 'Diagonal orientation'. The secret message/image bits stream in varying positions are embedded in 'Vertical orientation sub-image' followed by reverse transformation to generate embedded/encrypted image. The decoding is done through the reverse procedure. The experimental results against statistical and visual attack has been computed and compared with the existing technique like IAFDDFTT[1], in terms of Mean Square Error (MSE), Peak Signal to Noise Ratio (PSNR), Standard Deviation(SD) and Image Fidelity(IF) analysis, which shows better performances in ATFDWT.