Here we talk about image processing, laser marking machine, the principle is based on the standard controller and image processing theory, research and development of semiconductor laser for BMP image file format suitable for scanning marking laser marking software. The main task of the software is the more mature image processing method applied to the laser marking machine, get better picture quality, the system subsequent marking operation provides image data. Thesis emphasis on developing its own image processing software, the software includes two parts, namely, the image processing section and the implementation of marking parts. A BMP bitmap file typically includes a bitmap file header, bitmap header, palettes, bitmap data into four parts, based on Visual C ++ object-oriented programming methods to construct the device-independent bitmap CDIB library, complete the DIB file read writes and palette and the image display operation, providing the bitmap width, height, color, number and other relevant information. According to the marking requirements of image, color image gradation processing by the image gray-scale transformation, changing the original image occupied by the grayscale range to achieve a good visual effect. Geometric transformations to change the spatial relationship between the image of each object, a combination of image translation, mirroring, transposition, inversion, rotation processing, can achieve a variety of image geometric distortion, to achieve the object geometry and different geometry shape requirements. For laser marking systems, only two laser state of its control, namely the light and not light, the corresponding image is black and white, and therefore requires an image by thresholding to obtain binary image, depending on the image selection of appropriate threshold to achieve gray image binarization process. First detected edge detection image local characteristic discontinuities, and then not even then successive edge pixels into a complete boundary, to obtain a continuous image, a different edge detection operator image feature extraction for different images. Image formation, transmission, reception and processing of the process, will produce noise, image noise removal process can reduce the redundant information of the image. By erosion and dilation can better remove noise binary image. The refinement of the image highlights the shape characteristics and reduces redundant information. Structures based PC, D / A converter and dual galvanometer scanning system, laser marking machine control system, when marking, through D / A converter converts the image coordinate data into an analog voltage value, controls the rotation galvanometer to determine the laser light path and set a delay. Finally, here it describes the system hardware and software algorithms to achieve, given the image software program and debug marking results.