Due to a CCD and ADC restriction

the dynamic range of the commonly used image sensor is only about 60 dB

which does not coincide with the dynamic range(110 dB)of eyes

leading to the loss of partial details of the image. In this paper

an image extended dynamic range technique based on the wavelet transform would be proposed. By means of taking pictures of the same scene repeatedly

the source images of different exposure time were procured

which were further divided into high frequency part and the low frequency part by using multi-scale wavelet transform. The method based on human visual gray recognition features was applied to the processing of low frequency part. By using different weights for processing

the most background information possible of the image was retained. For the high frequency part

the local maximal variance criteria for fusion was used to maintain the image edges’ characteristics to the greatest extent. Reconstruction is proceeded by using inverse wavelet transform to the merged high frequency part and low frequency part. Test results showed that comparing the reconstructive image and moderately exposed image

the details of information in the high brightness area and low light area was recovered

and the dynamic range got significant expansion.