In this paper, the image steganography based on LSB and pixel classification is reviewed. Then, the method for steganography information in image is presented. This method based on LSB. Our purpose of this paper is to minimize the changes in cover image. At the first, the pixels of the image are selected to hiding the message; second complemented message will be hidden in LSB of selected pixels. In this paper, to solve some problems LSB method and minimize the changes, pixels categorized based on values of bits of second, third, fourth. In each category, ratio of changed pixels to unchanged pixels is calculated. If the ratio is greater than one, the LSB of that category are reversed and those changes reach at least. Mean Square Error and Peak Signal to Noise Ratio are two criterions to evaluate stego-image quality. PSNR and MSE of proposed method in comparison with simple LSB method, are respectively growth rate 0.13 percent and reduction rate 0.19 percent.