Minol Reader

Buildings using district heating often have a heat meter to measure the amount of heat used by the building. The Minol heat meter is a common heat meter used in Germany. It usually can display the current heat usage value on a seven segment display. In order to digitally capture the current reading of the Minol meter a custom system has been implemented: the Minol Reader.
The reading is then sent via MQTT to a Home Assistant instance. (Although with some modification the number without the use of MQTT or Home Assistant can be extracted)

The Minol Reader consists of:

• ESP Cam (ESP32-CAM) mounted onto the Minol meter
• Python script running on a separate PC or VM

Function

The ESP Cam is programmed to take a picture of the Minol meter every 60 minutes. The picture is then sent to the Python script processes the image to extract the current reading of the Minol meter with a simple seven segment segment optical character recognition (SSOCR) script adapted from this repository. A very simple calibration the display position is done via 2 calibration dots. The extracted reading is then sent via MQTT to a Home Assistant instance.

Installation

For the installation of the Python script Docker Compose and Docker are used.

Here are the steps to install the Minol Reader and the ESP Cam:

1. Clone this repository
2. Navigate to the repository
3. Setup ESP Cam with the CameraWebServer.ino sketch, if it is not already set up already. (Don't forget to change the ssid and password to match your WiFi network)
4. Edit the config.json file to match your MQTT broker and settings
5. Run docker-compose up -d to start the container

Configuration

The configuration of the Minol Reader is done via the config.json file. The file contains the following settings:

General settings:

Keyword	Type	Example	Description
version	str	"1.1"	Version String
esp_ip	str	---	IP of the ESP Cam
delay	int	60	Delay between image captures in minutes
save_images	bool	true	Save images to disk; can be used to verify; warning if images are not manually removed a lot of diskspace is used
monotone_detection	bool	true	Check if consecutive values are monotonicly rising; writes discrepancies to log file
exit_on_error	bool	true	Exit the script if an error occurs
mqtt_settings	dict	---	MQTT settings (explained later)
image_parameters	dict	---	Image processing parameters (explained later)

MQTT settings:

Keyword	Type	Example	Description
broker	str	---	IP of the MQTT broker
username	str	---	Username for the MQTT broker
password	str	---	Password for the MQTT broker
prefix	str	---	Prefix for the MQTT topic
topic_base	str	---	Base topic for the MQTT topic

Image parameters:

Keyword	Type	Example	Description
cut_indices	dict	{ "upper_left": [400, 305], "lower_right": [854, 415] }	Indices for the image cut taken from full image; contains "upper_left" and "lower_right"; both are lists with [x,y]
calib_centers	dict	{ "c1": [151, 499], "c2": [1175, 644] }	Centers for the calibration dots taken from full image; contains "c1" (left point) and "c2" (right point); both are lists with [x,y]
calib_areas	dict	{ "a1": { "upper_left": [100, 460], "lower_right": [200, 554] }, "a2": { "upper_left": [1120, 570], "lower_right": [1243, 701] } }	Areas for the calibration dots taken from full image; contains "a1" (left area) and "a2" (right area); both are defined by a dict containing "upper_left" and "lower_right" with format [x,y]
x_skew_factor	float	-0.06	Skew factor for the x-axis; used to correct the skew of the image
bi_end_left	int	125	Brightness correction gradient; end of left
bi_start_right	int	260	Brightness correction gradient; start of right
bi_max_left	int	75	Brightness correction gradient; max of left
bi_max_right	int	50	Brightness correction gradient; max of right
contrast_factor	int	2	Contrast factor for the image; used to increase the contrast of the image
skip_position_calibration	bool	false	Skip the position calibration; used to skip the calibration of the display position

Note:

• The brightness correction gradient is used to correct the brightness of the image. This is due to the LED of the camera, which more strongly illuminates the center of the image. A gradient on both sides of this center is used to correct this.

The coordinates used for the cut_indices, calib_centers and calib_areas can be found by using the calibrate_minol.py script and reading the indices from the full images.

Scripts

The repository contains the following scripts:

• minol_main.py: Main script for the Minol Reader (Started with Docker Compose)
• test_reading.py: Test reading script for the Minol Reader
• calibrate_minol.py: Calibration script for the Minol Reader (Images are shown in UI! Run on system with UI!)

Calibration

For the calibration run the calibrate_minol.py script.

The following images will be displayed when running the script:

1. Full image
2. Cut image
3. Cut image after skew correction, brightness correction and contrast correction

Getting correct calibration positions

To set the coordinates of the cut_indices, calib_centers and calib_areas the following steps are done:

1. Run the calibrate_minol.py script
2. Look at the Full image
3. Set the cut_indices to the area where the digits are displayed. Cut as tight as possible with only room to the left to allow for more digits. The area is defined by 2 points marking the upper left and lower right corner of the area.
4. Set the calib_centers to the center of the calibration dots.
5. Set the calib_areas to the area of the calibration dots. You can select a generous area around the calibration dots but leave enough room to not include any other dark surfaces. Allow for "wiggle room" if the camera moves slightly. The areas are defined by 2 points marking the upper left and lower right corners of the areas.
6. Set the other parameters as needed.
7. Rerun the calibrate_minol.py script to verify the settings in the cut image and the cut image after corrections.