ADD file via upload

blur_3.c

@@ -0,0 +1,110 @@
+#include <stdio.h>
+#include <time.h>
+#include <arm_neon.h>
+
+float kx[3] = { 0.25, 0.5, 0.25 };
+float ky[3] = { 0.25, 0.5, 0.25 };
+
+void applySeparableGaussianBlur_NEON(float src[5][5], float dst[5][5], int h, int w, float kx[3], float ky[3]) {
+    float buf[5][5] = { 0 };
+
+    float32x4_t kx_vec = vld1q_f32(kx);
+
+    for (int i = 0; i < h; i++) {
+        for (int j = 1; j < w - 1; j += 4) {
+            int end = (j + 4 < w) ? j + 4 : w;
+            for (int k = j; k < end; k += 4) {
+                if (k + 3 < w) {
+
+                    float32x4_t left = vld1q_f32(&src[i][k - 1]);
+                    float32x4_t mid = vld1q_f32(&src[i][k]);
+                    float32x4_t right = vld1q_f32(&src[i][k + 1]);
+
+
+                    float32x4_t result = vmulq_lane_f32(left, vget_low_f32(kx_vec), 0);
+                    result = vmlaq_lane_f32(result, mid, vget_low_f32(kx_vec), 1);
+                    result = vmlaq_lane_f32(result, right, vget_high_f32(kx_vec), 0);
+
+
+                    vst1q_f32(&buf[i][k], result);
+                }
+                else {
+
+                    for (int m = k; m < w && m < k + 4; m++) {
+                        buf[i][m] = src[i][m - 1] * kx[0] + src[i][m] * kx[1] + src[i][m + 1] * kx[2];
+                    }
+                }
+            }
+        }
+    }
+
+
+    float32x4_t ky_vec = vld1q_f32(ky);
+
+    for (int i = 1; i < h - 1; i++) {
+        for (int j = 0; j < w; j++) {
+            if (j < w - 1) {
+                if (i + 2 < h) {
+
+                    float32x4_t top = vld1q_f32(&buf[i - 1][j]);
+                    float32x4_t mid = vld1q_f32(&buf[i][j]);
+                    float32x4_t bottom = vld1q_f32(&buf[i + 1][j]);
+
+
+                    float32x4_t result = vmulq_lane_f32(top, vget_low_f32(ky_vec), 0);
+                    result = vmlaq_lane_f32(result, mid, vget_low_f32(ky_vec), 1);
+                    result = vmlaq_lane_f32(result, bottom, vget_high_f32(ky_vec), 0);
+
+
+                    vst1q_f32(&dst[i][j], result);
+                }
+                else {
+
+                    dst[i][j] = buf[i - 1][j] * ky[0] + buf[i][j] * ky[1] + buf[i + 1][j] * ky[2];
+                }
+            }
+            else {
+
+                dst[i][w - 1] = buf[i][w - 1];
+            }
+        }
+    }
+
+    for (int i = 0; i < h; i++) {
+        dst[i][0] = 0;
+        dst[i][w - 1] = 0;
+    }
+    for (int j = 0; j < w; j++) {
+        dst[0][j] = 0;
+        dst[h - 1][j] = 0;
+    }
+}
+
+int main() {
+    float src[5][5] __attribute__((aligned(16))) = {
+        {1, 2, 3, 4, 5},
+        {6, 7, 8, 9, 10},
+        {11, 12, 13, 14, 15},
+        {16, 17, 18, 19, 20},
+        {21, 22, 23, 24, 25}
+    };
+
+    float dst[5][5] = { 0 };
+
+    clock_t start = clock();
+    applySeparableGaussianBlur_NEON(src, dst, 5, 5, kx, ky);
+    clock_t end = clock();
+
+    printf("模糊后的图像矩阵:\n");
+    for (int i = 0; i < 5; i++) {
+        for (int j = 0; j < 5; j++) {
+            printf("%.2f ", dst[i][j]);
+        }
+        printf("\n");
+    }
+
+    double time_taken = (double)(end - start) / CLOCKS_PER_SEC;
+    printf("运行时间: %e 秒\n", time_taken);
+
+    return 0;
+}