Research On Salt - And - Pepper Noise Filtering Algorithm For Color Image Based On Uniform Spatial Color Difference Measurement

Before the work of converter flame recognition in metallurgical industry, the salt & pepper noise usually formed in flame image because of the smoke interference in the optical path or electronic components inside the cameral. The noise can reduce the quality of the image, and affect the subsequent recognition task. It is important to maintain the image information originally in the late period of flame image recognition. Therefore, this paper aimed to research the salt & pepper filtering algorithm for color images.The existing noise remove algorithms for color images are mainly focused on how to separate the noise more clearly from the RGB space. Though these algorithms played a vital role in the promotion of the image processing research, the algorithms cannot simultaneously get satisfactory results both in signal to noise ratio and color retention. In this paper, a salt & pepper noise removal algorithm for color image based on difference formula in uniform space is proposed, which can effectively remove the salt & pepper noise in color image without adding another information that beyond the original color image. Specific studies are as follows:Firstly, this paper combining the practical application background of color image filtering with some factors that may influence the result and comparing the shortcomings of existing classic color image filtering algorithm with proposed algorithm.Secondly, a color image de-noising algorithm based on uniform color space metric is proposed. The algorithm converts RGB space to uniform color space L*a*b*, makes the color difference is coordinated to the human eye perception, then sorting the color by its vector and removing the noise by the algorithm presented in this paper. Considering the deficiency of the original algorithm and in order to keep the details of the image and the de-noising result, proposed a uniform color space filtering algorithm with adaptive window. The experimental result shows that the present algorithm works well when the noise density is high.Finally, in order to apply the proposed algorithm to practice, algorithm experimental simulation system is developed in QT software platform based on Linux operating system. The result of image filtering proposed in this paper is verified in the experimental system; moreover, it can also compare the existed classical algorithms with this paper.The algorithm proposed in this paper not only obtains satisficing results, but also provides a new way to color image de-noising, which has good application and reference value.