Three-dimensional surface reconstruction of two-dimensional veterinary B-ultrasound images can more accurately and vividly display the pathological structures of various tissues and organs inside the animal body, and can quickly provide diagnostic information for veterinarians. However, compared with other medical images such as veterinary CT, veterinary X-ray and nuclear magnetic resonance, veterinary B-ultrasound images have poor resolution, few grayscale levels, narrow grayscale distribution range, certain noise interference, and no obvious grayscale changes between tissues and organs. The edges extracted by traditional methods are either too wide or intermittent, and cannot accurately and completely express the morphology of organs in veterinary B-ultrasound images. Therefore, they cannot be used as the basis for three-dimensional reconstruction. This is caused by the serious noise interference and incomplete contours of veterinary B-ultrasound images. First, the objects to be reconstructed in some typical veterinary B-ultrasound sequence cross-sectional images after edge enhancement are erased and filled, and only the single-pixel contour of the reconstructed object is extracted. Then, based on the edge detection method, an accurate semi-automatic contour extraction method for veterinary B-ultrasound image sequences for three-dimensional reconstruction is proposed.

Before contour extraction, preprocessing must be performed. The task of veterinary B-ultrasound image preprocessing is to suppress noise, enhance details, improve image quality, and provide a high-quality image with a high signal-to-noise ratio for subsequent processing such as feature extraction. Veterinary B-ultrasound images have shortcomings such as large noise, few grayscales, and poor contrast, and often have echo loss and other phenomena, resulting in incomplete contours. Therefore, high requirements are put forward for preprocessing, and a large number of algorithms have been tried. For different types of images, the same algorithm has different processing effects, and the effect of preprocessing the same veterinary B-ultrasound image is not only related to the algorithm, but also closely related to the order of processing. However, most of the existing image preprocessing algorithms do not consider the requirements of subsequent processing, and the effect of veterinary B-ultrasound image processing is not obvious. Therefore, based on the existing algorithms, transformation and innovation are carried out. After repeated analysis, a preprocessing method more suitable for veterinary B-ultrasound images is found. In actual application, the method adopted is to perform edge-maintaining smoothing filtering on the original veterinary B-ultrasound image, and then perform histogram equalization enhancement, which can well complete the task of image preprocessing. This processing method can provide a good digital image for contour extraction.

link: https://www.bxlimage.com/nw/188.html