Halcon机器视觉校准报错9001：修复方法及RGB图像转换

问题描述 #

使用Halcon开发机器视觉校准程序，通过图像采集助手连接Zivid外部相机并插入生成的采集代码后，调用find_calib_object时触发报错：

Image to process has wrong gray value type (HALCON error code: 9001)

同时需要实现将采集到的图像转为RGB格式的需求。

用户原始代码片段：

write_string(WindowHandle, 'Generating a position and sending it to the robot...')
* Using procedure to generate the position and send the robot
PositionGeneration (Socket, WindowHandle, TupleWithPositions)

* Image Acquisition 01: Code generated by Image Acquisition 01
open_framegrabber ('GenICamTL', 0, 0, 0, 0, 0, 0, 'default', -1, 'default', -1, 'false', 'default', 'Zivid_DEV_21513D73', 0, -1, AcqHandle)
grab_image_start (AcqHandle, -1)
grab_image_async (Image, AcqHandle, -1)
* Image Acquisition 01: Do something
close_framegrabber (AcqHandle)

dev_display (Image)

* Saving the picture
write_image (Image, 'png none', 0, 'Image'+I)

find_calib_object (Image, CalibDataID, 0, 0, I, 'alpha', 0.2)

get_calib_data_observ_contours (Caltab, CalibDataID, 'caltab', 0, 0, I)

get_calib_data_observ_points (CalibDataID, 0, 0, I, Row, Column, Index, Pose)

dev_set_color ('green')
dev_display(Image)
dev_display (Caltab)

tuple_number (TupleWithPositions, PoseCoordinates)
X := PoseCoordinates[0]
Y := PoseCoordinates[1]
Z := PoseCoordinates[2]
A := PoseCoordinates[3]
B := PoseCoordinates[4]
C := PoseCoordinates[5]

create_pose (X, Y, Z, A, B, C, 'Rp+T', 'abg', 'point', Pose)
set_calib_data (CalibDataID, 'tool', I, 'tool_in_base_pose', Pose)

用户已尝试修改相机参数，但未解决问题。

解决方案 #

1. 修复错误代码9001 #

错误核心原因：find_calib_object仅支持单通道灰度图像作为输入，而Zivid相机采集的是3通道彩色图像，灰度值类型不匹配导致报错。

解决步骤：

• 在调用find_calib_object前，使用rgb1_to_gray算子将采集到的彩色图像转换为灰度图像；
• 后续校准相关操作均使用转换后的灰度图像。

2. 将采集图像转为RGB格式 #

Zivid相机通过GenICam采集的图像默认多为BGR格式，需转换为标准RGB：

• 使用trans_from_rgb算子，指定从'bgr'转换为'rgb'格式；
• 若不确定图像格式，可先调用get_image_type查看图像类型，再选择对应转换方式。

修改后的完整代码示例 #

write_string(WindowHandle, 'Generating a position and sending it to the robot...')
* Using procedure to generate the position and send the robot
PositionGeneration (Socket, WindowHandle, TupleWithPositions)

* Image Acquisition 01: Code generated by Image Acquisition 01
open_framegrabber ('GenICamTL', 0, 0, 0, 0, 0, 0, 'default', -1, 'default', -1, 'false', 'default', 'Zivid_DEV_21513D73', 0, -1, AcqHandle)
grab_image_start (AcqHandle, -1)
grab_image_async (Image, AcqHandle, -1)
* Image Acquisition 01: Do something
close_framegrabber (AcqHandle)

* 将采集的BGR图像转换为标准RGB格式
trans_from_rgb(Image, ImageRGB, 'bgr', 'rgb')

* 将RGB图像转换为灰度图像，用于校准算子
rgb1_to_gray(ImageRGB, ImageGray)

dev_display (ImageRGB)

* 保存RGB格式的图像
write_image (ImageRGB, 'png none', 0, 'Image'+I)

* 使用灰度图像调用校准算子，解决类型不匹配问题
find_calib_object (ImageGray, CalibDataID, 0, 0, I, 'alpha', 0.2)

get_calib_data_observ_contours (Caltab, CalibDataID, 'caltab', 0, 0, I)

get_calib_data_observ_points (CalibDataID, 0, 0, I, Row, Column, Index, Pose)

dev_set_color ('green')
dev_display(ImageRGB)
dev_display (Caltab)

tuple_number (TupleWithPositions, PoseCoordinates)
X := PoseCoordinates[0]
Y := PoseCoordinates[1]
Z := PoseCoordinates[2]
A := PoseCoordinates[3]
B := PoseCoordinates[4]
C := PoseCoordinates[5]

create_pose (X, Y, Z, A, B, C, 'Rp+T', 'abg', 'point', Pose)
set_calib_data (CalibDataID, 'tool', I, 'tool_in_base_pose', Pose)

内容的提问来源于stack exchange，提问作者Viktor Petrov