Study For Impulse Noise Removal From Color Image

During the process of image acquisition, transmission, and storage, lots of noise are inevitably produced due to the influence of environmental factors, for example, the defects of imaging systems, transmission medium, and storage devices, etc. Most of these noises are Visible to human eyes and they exert great influences on the effect and accuracy of some subsequent image processing, such as edge detection, image segmentation, feature extraction, and pattern recognition, etc.Therefore, it is essential to restore noisy pixels before any subsequent image processing task is performed.In this paper, the current situation of noise removal and the basic relative knowledge for noise removal are introduced in advance. After having this basic knowledge for denoising, some tentative research has been done to solve the two problems of preserving image information and suppressing pseudo color. Then based on the metric of color difference, an algorithm for evaluation of color image is also introduced. All works are as follows:1.Firstly, the current situation of impulse noise removal from color images is researched; secondly, a common model of impulse noise is introduced and that why it is used in this paper is also explained; then, the image quality evaluation criteria is discussed from both subjective and objective aspects;finaly, the Spiking Cortical Model and its working mechanism are introduced briefly, and the synchronous pulse burst property of SCM and the working principle of SCM in the procedure of detecting impulse noise are also described in details.2.Because of these problems:the large computational complexity of vector-based denoising method, the randomness of the impulse noise and the arbitrariness of image strurture, the concept of fuzzy metric is introduced. Based on this concept, an adaptive algorithm is proposed,which removes impulse noise from color images with SCM detector. By the experimental results, this algorithm can effectively suppress the impulse noise while preserving the image details.When the number of the contanminated pixels does not exceed half of the total image pixels, the visual effect of the images denoised by the algorithm fully demonstrates its advantages on the aspects of noise suppression and image information protection.3.For the suppression of the pseudo-color and the reduction of the computational complexity, it is a feasible method to combine the scalar-based algorithm and the vector-based algorithm to remove noise from color images. Based on the idea which is mentioned previously, a new algorithm for the removal of impulse noise is proposed. The algorithm firstly uses SCM to detect impulse noise; then, it just restores the detected noise and keeps other pixels unchanged. By a large number of experimental data,it shows that whether the image is contaminated in a heavy level or not, the proposed new algorithm demonstrates a good performance.4. The definition of quaternion is introduced and its two transforms are also described. They are used to metric the color difference between two color pixels and the rationality has been proved in theory. Then, some experiments have been done to support it.