使用DigitalMicrograph计算不同ROI总计数/强度结果异常求助

代码中的核心问题 #

• 选区获取逻辑错误：处理第二个ROI时，手动硬编码t2=0;l2=x0+1;b2=1;r2=xmax+1覆盖了正确的选区获取，且后续错误调用img1.GetSelection()，此时前台图像已因showimage(cropped)切换，导致选区信息完全不符合预期。
• 参数混用：计算第二个ROI的integral_cal2时，错误使用了第一个图像的scale参数，而非当前图像的scale2，导致校准结果偏差。
• 依赖前台图像不稳定：GetFrontImage()依赖当前激活窗口，操作中切换图像会导致获取的图像对象与目标不符，结果不可靠。

修正后的代码 #

ROI roi_2 = NewROI()
ROISetRange( roi_2, 10, 30 )
cropdisp.ImageDisplayAddROI( roi_2)
imagedisplaysetroiselected(cropdisp, roi_2,1)

number count_l, count_r
image img_source := GetFrontImage()  // 提前锁定源图像，避免依赖前台窗口切换

// 处理第一个ROI
number t,l,b,r
img_source.GetSelection(t,l,b,r)
result("\n select1 is"+t+","+l+","+b+","+r)
image roi_img_1 := img_source[t:b, l:r]  // 直接截取ROI区域图像
number integral = sum( roi_img_1[] )
number scale = img_source.ImageGetIntensityScale()
number origin = img_source.ImageGetIntensityOrigin()
number nChannels = r - l
number integral_cal = (integral - origin * nChannels) * scale
count_l= integral_cal
result("\n count_l= "+count_l +", ")

// 处理第二个ROI
ROI roi_3 = NewROI()
ROISetRange( roi_3, 35, 55 )
cropdisp.ImageDisplayAddROI( roi_3)
imagedisplaysetroiselected(cropdisp, roi_3,1)

img_source.GetSelection(t,l,b,r)  // 从锁定的源图像获取最新选区
result("\n select2 is"+t+","+l+","+b+","+r)
image roi_img_2 := img_source[t:b, l:r]
number integral2 = sum( roi_img_2[] )
number scale2 = img_source.ImageGetIntensityScale()
number origin2 = img_source.ImageGetIntensityOrigin()
number nChannels2 = r - l
number integral_cal2 = (integral2 - origin2 * nChannels2) * scale2
count_r= integral_cal2
result("\n count_r= "+count_r +", ")

更优实现思路：直接基于ROI掩码计算 #

无需依赖选区，直接通过ROI生成掩码图像计算目标区域强度，彻底避免窗口切换带来的问题：

image img_source := GetFrontImage()
imageDisplay disp = img_source.ImageGetImageDisplay(0)

// 计算左侧ROI
ROI roi_left = NewROI()
ROISetRange( roi_left, 10, 30 )
disp.ImageDisplayAddROI( roi_left )
image mask_left = roi_left.ROIGetMaskImage(img_source)
number count_l = sum( img_source * mask_left )
// 应用强度校准
number scale = img_source.ImageGetIntensityScale()
number origin = img_source.ImageGetIntensityOrigin()
count_l = (count_l - origin * sum(mask_left)) * scale
result("\n count_l= "+count_l +", ")

// 计算右侧ROI
ROI roi_right = NewROI()
ROISetRange( roi_right, 35, 55 )
disp.ImageDisplayAddROI( roi_right )
image mask_right = roi_right.ROIGetMaskImage(img_source)
number count_r = sum( img_source * mask_right )
count_r = (count_r - origin * sum(mask_right)) * scale
result("\n count_r= "+count_r +", ")

内容的提问来源于stack exchange，提问作者together