Image Enhancement Algorithm Based On Homomorphic Filtering

Digital camera while acquiring the image, often because of the impact of complex light environment, leading to the appearance of the image obtained insufficient illumination, the image is dim, picture clarity is not high.Image Enhancement image processing technology as a basic processing techniques, the use of airspace in the frequency domain or the image analysis and processing, improve the image quality can be very good to make the image clearer, more prominent details, the visual effect is more in line with the human eye visual characteristics.Where in the frequency domain based illumination-homomorphic reflection model enhancement method in dealing with low-light image, the image can be avoided direct Fourier transform to produce distortion, while maintaining the original image, a good image to enhance dark areas details, eliminating the problem of uneven illumination.Usually people are most exposed to a color image, color image are more gray image contains more information, more conducive to human-computer observation and analysis identified. Therefore, the study and improvement of methods of color image enhancement processing has more significance.Therefore, this paper,based on image enhancement method homomorphic filtering, primarily by improving the homomorphic filtering algorithm, to achieve the illumination color enhanced image brightness and contrast, making it clearer, more obvious details. The main work is as follows:First, the traditional homomorphic filtering algorithm is studied and discussed, starting from the transfer function, design without changing the image enhancement processing effects, and less easy to adjust the parameter content of the new transfer function, which improved homomorphic filtering can quickly realize enhanced image detail and brightness.Second, the use of the limited contrast histogram equalization (CLAHE) algorithm can enhance image detail and balanced image gradation characteristics of the improved homomorphic filtering algorithm in combination with, proposed a new algorithm that can not only enhance the image of brightness and contrast to make the image clearer, and highlights the more details, the image contains more information and is more suitable for human-computer analysis of observation. And through multiple sets of experiments, to verify the effectiveness of the algorithm and has enhanced low light color image at the optimal treatment effect.Third, in order to obtain higher quality color images, the above method to improve again, we propose two homomorphic filtering color image processing method based on color space conversion. The first stretch of the luminance homomorphic filtering based on color image enhancement algorithm, the algorithm uses homomorphic filtering for improved RGB color image enhancement for the first time, then, will the enhanced image into HSV color space, holding chrominance components H constant luminance component V limited adaptive histogram equalization and contrast piecewise linear transformation, and the saturation S doing stretching 1.5 times, enhance saturation of the image. Finally, the HSV color space is converted back to RGB color space, thus completing the whole process enhancement. The second is a nonlinear homomorphic filtering color image enhancement based on illumination compensation algorithm through improved by homomorphic filter algorithm and CLAHE algorithm processed image into HSV space, keeping constant chromaticity, using the constructed V as a function of the luminance component processing, illumination compensation, and saturation to conduct a 1.5-fold stretching. Finally, the brightness, saturation of the original color and the enhanced information fusion and restore the image color information to obtain the final enhanced image.Multi-set of experiments to verify the two algorithms, experiments show that the two algorithms have the effect of enhanced image clarity and detail, and the enhanced image full color, more in line with human visual perception characteristics.