A Wavelet Neural Network-based Low-contrast Image Enhancement Method

As the military and industry is developing dramatically, to enhance the low-contrast digital image has become one of the important areas in image processing. There are two traditional methods to enhance the low-contrast digital image: the methods based on frequency and the method based on time. The histogram enhancement, the most popular way to enhance, is representing the latter one; while the methods based on frequency often uses wavelet transform, FT(flourier transform), DCT(discrete cosine transform) to enhance. This research mainly discusses the wavelet neural network enhancement of the methods based on frequency.The wavelet has good transforming character, and the wavelet neural network makes good use of this character, to make the image signal resolve sufficiently in term of frequency after the wavelet transform, and separate the information to reflect the outline and details of the image. This method enhances the transformed wavelet coefficient, strengthen the wavelet coefficient which reflects the outline of the image, while weaken the one which reflects the details of the image. Then it transforms the wavelet coefficient in a reverse way, therefore to get the enhanced image.There are two issues in this processing. One issue is the selection of the wavelet coefficient. The resulting wavelet coefficients after the transform are greatly different if we choose different wavelet coefficients to deal with the image. Of course, the information reflecting the outline of the image is also different. That is why to chose the most suitable wavelet coefficient has a crucial effect. Based on many project experience and laboratory experiment, we choose the Db44 wavelet.Another issue is the selection of the times of the wavelet coefficient. Our target choosing a parameter is to optimize the wavelet coefficient which reflects the outline of the image. To achieve this target, we introduce the neural network, and use the BP neural network to select and then get the best one. The neural network has a character of...