Research Single Image Defogging Algorithm Based On HSI Color Space

Nowadays, the image information has led to a very range of knowledge in the field. Digital image processing is a common tool for people to acquire, process and optimization, and the computer vision system is a more challenging application because of it intersects with the digital image processing and other subjects. However, along with the increase in haze image, reducing the effect of haze weather conditions on the outdoor visual equipment, will be taken by fog blurred, low visibility, lower contrast and color degradation of image processing into a high visibility, high resolution and colorful image, and to quickly and efficiently to the image processing, it is urgent need of reality and theory.In order to solve the problem of image contrast mass reduction, this paper proposes a defogging algorithm of image based on HSI color space. Firstly, in order to the hue of approximate invariance in the defogging around, the saturation value of the no fog bright image tends to a maximum of 1, and the intensity value becomes larger as the concentration of the fog is enhanced as the basic facts. Secondly, based on the above facts and the size relationship of the corresponding component in the defogging around, we can establish the linear defogging model which meets their respective characteristics. Then based on the theory that saturation value of the bright image without fog tends to a maximum of 1, it works out the component T(x) in the saturation defogging model. And for the component t(x) in the intensity defogging model is through its own characteristics to determine the value range, then make a appropriate approximation, and the approximation value results show that the component t(x) is related to the maximum value of the saturation in the selected area. Finally, the effective recovery of a clear image.The paper mainly made the following tasks:(1) By analyzing the characteristics of the three components in the HSI color space can be obtained that the tone H has approximate invariance after defogging, and combined with the saturation S and the intensity I to meet the size relationship in defogging around, to establish the linear defogging model.(2) Combined with the characteristics of the saturation value in the saturation component image without fog obtained the theoretical basis that in the most regions, the image without fog can satisfy the formula:max Ω(x)(S2 (x)) →1. Then there is a linear estimation for the special image of parameters in the intensity model to establish a linear defogging model.(3) By using inequality relation to improve the intensity model of the general image, combined with their own characteristics of the parameters in the intensity model, the value range of the parameters can be determined, and then make a appropriate approximation, the approximation results show that the parameters related to the maximum value of the saturation of the selected area. Finally, the effective recovery of a clear image.In order to verify the validity of the algorithm, this paper was tested with the VC++6.0 programming. Experiments shows that the algorithm removes some fog in close range area and maintains the authenticity of the scene, at the same time highlighted landscape details in the fog blurred vision area, and the hue compared with the original image is little difference, and the whole image looks more natural. Then by compared experimental results with other algorithms from the visual effects and computing speed, it is show that the running efficiency of the method is more than doubled, and the proposed method effectively enhanced clarity, so it is appropriate for single image defogging.