| The infrared imaging technology is drawing more and more attention in the world, and it's application is more in the night vision system. Due to the influence of the atmosphere, the objects'inherent infrared characteristic and other factors, infrared image has disadvantages of low contrast between the target and background, blurry edge and high noise. So the infrared image must be enhanced. Among the existing infrared image enhancement algorithms can enhance the gray contrast of the image, but it depresses the visual effectiveness with the noise excessively enhanced.The result of image processing is often evaluated by human eyes. So it's necessary to process the captured infrared image to make the image more suitable for human visual system. Then better visual effectiveness can be obtained and the enhanced infrared image to identify the faint objectives. The characteristic of infrared image is analyzed. First of all, characteristics of the infrared image are researched, including histograms, contrast and noise. The histograms of infrared image are analyzed and compared with histograms of visible light image. The important question is concluded for enhancing infrared image. Then the human visual system (HVS) is researched, including contrast sensitivity, brightness and gray-sensitive properties, space frequency characteristics, to cover up the visual characteristics of the image of the Mach band effect, and so on. The human eye in the perception of natural scene, attention always focuses on the inerratic fragmentation and feature contour of object shape, and neglects general smooth transition region. The fractal characteristics of infrared image are analyzed by fractal theory and enhancement method is proposed based on HVS.After analyzing the characteristics of infrared images and human visual system, the real sensitive information of the human eye are specifically enhanced in the enhancement. In this way spatial frequency characteristics and multi-channel features and sensitive to the gray image of the human eye are used to enhance the details and contrast of image. The fractal and wavelet as tools are used to extract the sensitive part of the human eye and enhance the sensitive & non-sensitive areas in the image, the better visual effect than the conventional method of infrared image will be gotten. In this paper, based on the analysis of characteristics of infrared image, algorithms for infrared image of enhancement are proposed. The important research works of the dissertations include follow: 1. Enhancement method of infrared image using fractal based HVS proposed for elimination of blurred edge in the fourth chapter. Fractional Brownian motion (FBM) is one of the fractal theory used for evaluation of image in a varying range of gray levels. One important step in the FBM procedure is the measurement of Hurst exponent (H), which is directly related to the fractal dimension and surface roughness of a natural texture. The fractal dimension of a texture and the roughness of the image surface fit closely with the visual perception of human eyes. HVS is sensitive to the edge area of the image, so the pixels are classified into edge pixels, smooth pixels and texture pixels by their fractal dimensions. The gray level of each edge pixels are weighted and enhanced using a fractal based HVS.2. A multi-fractal theory analysis of infrared image is proposed and the multi-fractal characteristics of edge in the infrared image are extracted. A novel multi-fractal analysis approach to the problem of image edge detection is proposed in this paper. The singularity characteristic of each pixel in the image and its multi-fractal spectrum is calculated. Then pixels are classified by human visual system (HVS), which is more sensitive to the edge structure of image. From smoothness area to edge area, the pixels are weighted and enhanced. Experiments of the image enhancement method are taken and results of experiments are compared with results of histogram equalization.3. A Retinex theory is combined with wavelet algorithm for infrared image enhancement. Wavelet de-noising and edge detection are merged into image enhancement, that is, the image de-noising and enhancement are accomplished at the same time. After the image is decomposed, its edges, details and noise will exist in high frequency. The de-noising and enhancement are linked by enhancement function. The image can be de-noised and enhanced by adjusting the parameters of enhancement function. Under the influence of infrared detecting instrument and outside lightness, the unwanted luminance drop and halo will appear. Infrared image often appear uneven illumination, too bright or partial halo. The low-frequency information of Wavelet is enhanced by MSR algorithm to improve contrast and reduce nonuniformity of the image.Finally, the three image enhancement methods based on human eye are introduced into image processing unit. The main design idea of monitoring system is expatiated. Experimental results have indicated that enhancement method based on human eye is effective on detecting the edge detail of infrared image. This approach can be used to preserve edge details and enhance the contrast of image.
|
PDF Full Text Request