Optimize Python script for generating Point Cloud

I'm trying to produce a 3D point cloud from a depth image and some camera intrinsics. The image is 640x480, and is a NumPy array of bytes. The output is a (rows * columns) x 3 array of points.

I've gotten the function to work perfectly, but it's way too slow! (takes like 2 seconds per image to process). I'm wondering if there are any optimizations I can make before giving up and writing a C module.

def create_point_cloud(self, depth_image):
    shape = depth_image.shape;
    rows = shape[0];
    cols = shape[1];

    points = np.zeros((rows * cols, 3), np.float32);

    bytes_to_units = (1.0 / 256.0);

    # Linear iterator for convenience
    i = 0
    # For each pixel in the image...
    for r in xrange(0, rows):
        for c in xrange(0, cols):
            # Get the depth in bytes
            depth = depth_image[r, c, 0];

            # If the depth is 0x0 or 0xFF, its invalid.
            # By convention it should be replaced by a NaN depth.
            if(depth > 0 and depth < 255):
                # The true depth of the pixel in units
                z = depth * bytes_to_units;

                # Get the x, y, z coordinates in units of the pixel
                points[i, 0] = (c - self.cx) / self.fx * z;
                points[i, 1] = (r - self.cy) / self.fy * z;
                points[i, 2] = z
            else:
                # Invalid points have a NaN depth
                points[i, 2] = np.nan;
            i = i + 1
    return points