您现在的位置是：首页 >技术教程 >C# - 图片抠像 - U2NET - 含模型 - 完整可运行网站首页技术教程

环境

.NET Framework 4.6.2 ， X64 , VS2022

依赖 

OpenCvSharp.dll，OpenCvSharp.Extensions.dll，Microsoft.ML.OnnxRuntime.dll，u2net_human_seg.onnx

效果

页面

全部逻辑代码

    public partial class Form1 : Form
    {
        private string imgFile = "";  
        private string model_path = string.Empty;
        private int modelSize = 0;
        private SessionOptions options = null;
        private InferenceSession onnx_session = null;
        private List<NamedOnnxValue> input_ontainer = null;
        private DenseTensor<float> input_tensor = null;


        public Form1()
        {
            InitializeComponent();
        }

        private void button1_Click(object sender, EventArgs e)
        {
            using (OpenFileDialog ofd = new OpenFileDialog())
            {
                ofd.Filter = "图像文件|*.jpg;*.png;";
                if (ofd.ShowDialog() == DialogResult.OK)
                {
                    imgFile = ofd.FileName;
                }
            }
            if (string.IsNullOrEmpty(imgFile))
            {
                return;
            }

            do_rec();
        }

        private async void do_rec()
        {
            var imgBs = File.ReadAllBytes(imgFile);
            var imgInit = Image.FromStream(new MemoryStream(imgBs)) as Bitmap;
            pb_init.Image = imgInit;
            Application.DoEvents();


            if (input_tensor == null)
            {
                load_removebg();
            }

            var imgRst = await Task.Run(() =>
               removebg(imgBs)
            );
            pb_rst.Image = imgRst;
            Application.DoEvents();
        }

        private void load_removebg()
        {
            this.model_path = AppDomain.CurrentDomain.BaseDirectory + "\model\u2net_human_seg.onnx";

            modelSize = 320;

            //创建输出会话，用于输出模型读取信息
            options = new SessionOptions();
            options.LogSeverityLevel = OrtLoggingLevel.ORT_LOGGING_LEVEL_INFO;
            options.AppendExecutionProvider_CPU(0);
            onnx_session = new InferenceSession(model_path, options);

            input_ontainer = new List<NamedOnnxValue>();
            input_tensor = new DenseTensor<float>(new[] { 1, 3, modelSize, modelSize });

        }

        private unsafe Bitmap removebg(byte[] imgbs)
        {
            var image = Cv2.ImDecode(imgbs, ImreadModes.Color);

            int oldwidth = image.Cols;
            int oldheight = image.Rows;


            var lenMin = Math.Min(oldwidth, oldheight);
            image = image.Resize(new OpenCvSharp.Size(lenMin, lenMin));


            int maxEdge = Math.Max(image.Rows, image.Cols);
            float ratio = 1.0f * modelSize / maxEdge;
            int newHeight = (int)(image.Rows * ratio);
            int newWidth = (int)(image.Cols * ratio);
            Mat resize_image = image.Resize(new OpenCvSharp.Size(newWidth, newHeight));
            int width = resize_image.Cols;
            int height = resize_image.Rows;
            if (width != modelSize || height != modelSize)
            {
                resize_image = resize_image.CopyMakeBorder(0, modelSize - newHeight, 0, modelSize - newWidth, BorderTypes.Constant, new Scalar(255, 255, 255));
            }

            Cv2.CvtColor(resize_image, resize_image, ColorConversionCodes.BGR2RGB);

            for (int y = 0; y < resize_image.Height; y++)
            {
                for (int x = 0; x < resize_image.Width; x++)
                {
                    input_tensor[0, 0, y, x] = (resize_image.At<Vec3b>(y, x)[0] / 255f - 0.485f) / 0.229f;
                    input_tensor[0, 1, y, x] = (resize_image.At<Vec3b>(y, x)[1] / 255f - 0.456f) / 0.224f;
                    input_tensor[0, 2, y, x] = (resize_image.At<Vec3b>(y, x)[2] / 255f - 0.406f) / 0.225f;
                }
            }

            string inputName = onnx_session.InputMetadata.Keys.First();
            this.input_ontainer.Clear();
            input_ontainer.Add(NamedOnnxValue.CreateFromTensor(inputName, input_tensor));

            //运行 Inference 并获取结果
            var result_infer = onnx_session.Run(input_ontainer);

            //将输出结果转为DisposableNamedOnnxValue数组
            var results_onnxvalue = result_infer.ToArray();

            //读取第一个节点输出并转为Tensor数据
            var result_tensors = results_onnxvalue[0].AsTensor<float>();

            var result_array = result_tensors.ToArray();

            float maxVal = result_array.Max();
            float minVal = result_array.Min();
            for (int i = 0; i < result_array.Length; i++)
            {
                result_array[i] = (result_array[i] - minVal) / (maxVal - minVal) * 255;
            }
            var result_image = new Mat(modelSize, modelSize, MatType.CV_32F, result_array);

            Cv2.CvtColor(result_image, result_image, ColorConversionCodes.RGB2BGR);


            #region  图像合并  和  尺寸恢复


            var m1 = image;
            var m2 = result_image;

            m1 = m1.Resize(new OpenCvSharp.Size(oldwidth, oldheight));
            m2 = m2.Resize(new OpenCvSharp.Size(oldwidth, oldheight));


            using (var result = new Mat())
            {
                try
                {
                    // 首先将m2转换为单通道灰度图
                    using (var grayMat = new Mat())
                    using (var normalizedMat = new Mat())
                    {
                        // 如果m2是3通道，先转为灰度图
                        if (m2.Channels() == 3)
                        {
                            Cv2.CvtColor(m2, grayMat, ColorConversionCodes.BGR2GRAY);
                        }
                        else
                        {
                            m2.CopyTo(grayMat);
                        }

                        // 将32F转换为8U，并进行归一化
                        if (grayMat.Type() == MatType.CV_32F || grayMat.Type() == MatType.CV_32FC3)
                        {
                            // 归一化到0-255范围
                            Cv2.Normalize(grayMat, normalizedMat, 0, 255, NormTypes.MinMax);
                            normalizedMat.ConvertTo(grayMat, MatType.CV_8UC1);
                        }

                        // 转换m1的颜色空间从BGR到RGB
                        using (var rgbMat = new Mat())
                        {
                            Cv2.CvtColor(m1, rgbMat, ColorConversionCodes.BGR2RGB);

                            // 分离RGB通道
                            var channels = rgbMat.Split();

                            // 创建包含4个通道的列表：RGB + Alpha
                            var allChannels = new List<Mat>();

                            // 按照RGBA顺序添加通道
                            allChannels.Add(channels[2]); // R
                            allChannels.Add(channels[1]); // G
                            allChannels.Add(channels[0]); // B
                            allChannels.Add(grayMat.Clone()); // Alpha

                            // 合并所有通道，创建RGBA图像
                            Cv2.Merge(allChannels.ToArray(), result);

                            // 清理分离的通道
                            foreach (var channel in channels)
                            {
                                channel.Dispose();
                            }

                            // 清理通道列表
                            foreach (var channel in allChannels)
                            {
                                channel.Dispose();
                            }

                            // 转换为Bitmap
                            return OpenCvSharp.Extensions.BitmapConverter.ToBitmap(result);
                        }
                    }
                }
                catch (Exception ex)
                {
                    MessageBox.Show($"处理图像时出错: {ex.Message}
调试信息:
{ex.StackTrace}");
                    return null;
                }
            }
            #endregion

        }

    }

Demo 下载地址：

https://download.csdn.net/download/xuezhandansha/90357079