Motion Detection Settings

These are the options that controls the detection of motion. Further details follows after.

area_detect

• Type: String
• Range / Valid values: 1 - 999999999
• Default: Not defined
• Option Topic

Detect motion center in predefined areas. A script (on_area_detected) is started immediately when motion center is detected in one of the given areas, but only once during an event even if there is motion in a different configured area.

Areas are numbered like that:

      1    2    3
      4    5    6
      7    8    9

One or more areas can be specified with this option.

Example: You want to monitor if the center of motion occurrs in the lower third of the image - that is area 7, 8 and 9. Simply set 'area_detect' to '789' and 'on_area_detect' will be executed as soon as the center of motion was detected in area 7, 8 or 9. If you want to monitor area 2, 3, 5 and 6, set '2356'.

despeckle

• Type: String
• Range / Valid values: EedDl
• Default: Not defined
• Option Topic

Despeckle motion image using combinations of (E/e)rode or (D/d)ilate. And ending with optional (l)abeling.

A way of tuning (by removing or enhancing) noise in the motion image. Options for the despeckle feature are any of 'e', 'E', 'd' or 'D'. This can be combined by a trailing 'l' (letter l) which enables the labeling feature. Default: Not Defined (Don't despeckle and label).

Wind blowing grass and trees around or poor light conditions can cause a lot of dots (or noise) to appear in the motion image (See the section on Tuning Motion). This feature removes (or enhances!) this noise and so improves the reliability of motion.

The 'e' option removes diamonds, 'E' removes squares and alternating eE will remove circles. Each e/E you add will shrink the noise by a pixel all the way around. So 'despeckle Ee' will remove circles of radius 2. However, this will also shrink the detection by 2 and will affect the threshold. So to remove noise and then restore the detected motion to its original size try 'despeckle EedD'.

After the despeckle feature is done you can let the labeling feature search for areas of connected pixels and "label" each area. The program will now trigger motion based on the number of changed pixels in the largest area. In other words, the largest labeled area has to be above the threshold to trigger a motion detected.

The value EedDl is a good starting point. The possible combinations are endless and it requires many experiments to find the best combination. Just remember that the labeling feature only works as intended if it runs after the despeckle feature. Ie. the letter 'l' must be the last letter and only one 'l'.

If you have very few problems with false detections leave this option either blank or at EedD which will remove most of the single pixel noise. A very detailed technical explanation of the despeckle part can be found at the webpage of the author of this feature Ian McConnell's Webcam: Motion Web Page

gap

• Type: Integer
• Range / Valid values: 0 - 2147483647
• Default: 60
• Option Topic

Gap is the seconds of no motion detection that triggers the end of an event. An event is defined as a series of motion images taken within a short timeframe.

Recommended value is 60 seconds (Default). The value 0 is allowed (but not recommended) and disables events causing all Motion to be written to one single mpeg file and no pre_capture. You can force an event to end and a new to begin using the http control 'http://host:port/thread_number/action/makemovie'. Disabling events has bad side effects on noise_tune and smartmask. Both features can only work properly outside an event. When gap is set to 0, both features don't work properly anymore.

An event is defined as a series of motion images taken within a short timeframe. E.g. a person walking through the room is an event that may have caused 10 single jpg images to be stored. This option defines how long a pause between detected motions that is needed to be defined as a new event.

The gap timer starts after the last motion is detected and post_capture images have been saved and appended to open movie mpeg files.

Any motion detected before the gap timer times out resets the gap timer so it starts counting over again.

Detailed Description

The option 'gap' is important. It defines how long a period of no motion detected it takes before we say an event is over. An event is defined as a series of motion images taken within a short timeframe. E.g. a person walking through the room is an event that may have caused 10 single jpg images to be stored. Motion detected includes post_captured frames set by the 'post_capture' option. The 'gap' option defines how long a pause between detected motions that is needed to be defined as a new event. A good starting value is 60 seconds.

The way 'gap' works in more technical terms is:

• Gap is a timer that timeout 'gap' seconds after the last video frame with motion is detected.
• If 'post_capture' is activated then the gap timer starts counting after the last image of the post_capture buffer has been saved.
• The gap timer is reset and starts all over each time new motion is detected, so you will not miss any action by having a short 'gap' value. It will just create more events (e.g. more mpegs files)

The gap value impacts many functions in Motion.

• When the gap timer runs out the event number is increased by one next time motion is detected. When you use the %v conversion specifier in filenames or text features this means that the number in filename or text increased by one.
• The pre_capture feature only works at the beginning of an event. So if you have a very large 'gap' value pre_capture is not working very often.
• When you make mpegs using the ffmpeg features a new mpeg file is started at the beginning of an event when the first motion is detected. When 'gap' seconds has passed without motion (and post_captured frames saved) the mpeg files is completed and closed.
• Do not use large gap values to generate one large mpeg4 file. If Motion stops working this mpeg4 file never gets properly completed and closed and you will not be able to view it.
• Some of the tracking features sets the camera back to the center position when an event is over.

Note that 'gap' and 'minimum_gap' have nothing to do with each other.

lightswitch

• Type: Integer
• Range / Valid values: 0 - 100
• Default: 0 (disabled)
• Option Topic

Ignore sudden massive light intensity changes given as a percentage of the picture area that changed intensity.

Experiment to see what works best for your application.

Note: From version 3.1.17 (snap release 2 and on) this option has changed from a boolean (on or off) to a number in percent between 0 and 100. Zero means the option is disabled.

The value defines the picture areas in percent that will trigger the lightswitch condition. When lightswitch is detected motion detection is disabled for 5 picture frames. This is to avoid false detection when light conditions change and when a camera changes sensitivity at low light.

mask_file

• Type: String
• Range / Valid values: Max 4095 characters
• Default: Not defined
• Option Topic

PGM file to use as a sensitivity mask. This picture MUST have the same width and height as the frames being captured and be in binary format.

Full path of the PGM (portable gray map) mask file (binary format).

If you have one or more areas of the camera image in which you do NOT want motion detected (e.g. a tree that moves in the wind or a corner of the picture where you can see cars/pedestrians passing by) you need a mask file. This file is a picture that you create in your favorite photo editing program. The areas that you want detected must be white. The error that you want ignored must be black. The pgm image must be the same size (number of pixels high and wide) as the pictures that are taken by the camera (video4linux device).

You can adjust sensitivity by using gray tones.

If you do not have a mask file disable this option by not having it in the config file or comment it out ("#"or ";" as first character in line). If you are using the rotate option, note that the mask is applied after the rotation.

Detailed Description

The mask file must be a pgm format image file (portable gray map). Note that you must choose the BINARY format.

The feature is simple. Create an image of exact the same size as the ones you get from your video device (camera). Make a purely white picture and paint the areas that you want to mask out black. You can also make gray areas where you want to lower the sensitivity to motion. Normally you will stick to pure black and white.

One easy method for generating the mask file is as follows.

You can just take a motion captured picture, edit it with black and white for the mask and save it as a pgm file. If you cannot save in this format save as a grayscale jpg and then you can convert it to pgm format with

djpeg -grayscale -pnm [inputfile] > mask.pgm

(assuming you have djpeg installed - part of the jpeg lib package).

Note that the mask file option masks off the detection of motion. The entire picture is still shown on the picture. This means that you cannot use the feature to mask off an area that you do not want people to see.

Below are an example of a webcam picture and a mask file to prevent the detection cars in the street.

Normal picture. Notice the street is visible through the hedge.

Mask file (converted to png format so it can be shown by your web browser)

max_mpeg_time

• Type: Integer
• Range / Valid values: 0 (infinite) - 2147483647
• Default: 3600
• Option Topic

The maximum length of an mpeg movie in seconds. Set this to zero for unlimited length.

minimum_motion_frames

• Type: Boolean
• Range / Valid values: 1 - 1000s
• Default: 1
• Option Topic

Picture frames must contain motion at least the specified number of frames in a row before they are detected as true motion. At the default of 1, all motion is detected. Valid range is 1 to thousands, but it is recommended to keep it within 1-5.

Note that the picture frames are buffered by Motion and once motion is detected also the first frames containing motion are saved so you will not miss anything.

The feature is used when you get many false detections when the camera changes light sensitivity or light changes.

Experiment for best setting. Even though Motion accepts large values you should set this to a relatively low number (below 10). For each step larger than 1 Motion reserves space in RAM for the picture frame buffer. If you have a large value Motion will miss many frames from the camera while it is processing the all the pictures in the buffer.

noise_level

• Type: Integer
• Range / Valid values: 1 - 255
• Default: 32
• Option Topic

The noise level is used as a threshold for distinguishing between noise and motion.

This is different from the threshold parameter. This is changes at pixel level. The purpose is to eliminate the changes generated by electric noise in the camera. Especially in complete darkness you can see the noise as small grey dots that come randomly in the picture. This noise can create false motion detection. What this parameter means is that the intensity of a pixel must change more than +/- the noise threshold parameter to be counted.

noise_tune

• Type: Boolean
• Range / Valid values: on, off
• Default: on
• Option Topic

Activates the automatic tuning of noise level.

This feature makes Motion continuously adjust the noise threshold for distinguishing between noise and motion. The 'noise_level' setting is ignored when activating this feature. This is a new feature and new algorithm. It may give different results depending on camera and light conditions. Report your experience with it on the Motion mailing list. If it does not work well, deactivate the 'noise_tune' option and use the manual setting of 'noise_level' instead.

output_all

• Type: Boolean
• Range / Valid values: on, off
• Default: off
• Option Topic

Picture are saved continuously as if motion was detected all the time.

This feature is not meant to be the normal mode of operation. Especially not if you have the output_normal or output_motion features enabled since it will keep on saving pictures on the disk and you will soon run out of disk space. So be careful with this command.

If your frame rate is 10 pictures per second motion will save 10 new picture pr second until the disk is full.

It does all the normal actions that are done when motion is detected. It saves pictures on the harddisk, execute external scripts, etc as fast as the frame rate of the camera. So it is probably a good idea to run with a low framerate when using this feature and to not use activate all the features that saves files on the disk.

The idea of this feature is that you can turn the feature on and off for a short period of time to test or to generate continuous mpeg films when needed.

post_capture

• Type: Integer
• Range / Valid values: 0 - 2147483647
• Default: 0 (disabled)
• Option Topic

Specifies the number of frames to be captured after motion has been detected.

The purpose of this is mainly to create smooth video clips each time motion is detected. Use it to you personal taste (and disk space)..

This option is the preferred way to create continuous movies. Post_capture does not consume extra RAM and it does not create pauses in the movie even with large values.

If you only store mpegs movies and do not have output_normal on, then the recommended post_capture value is what is equivalent to 1-5 seconds of movie.

pre_capture

• Type: Integer
• Range / Valid values: 0 - 100s
• Default: 0 (disabled)
• Option Topic

Specifies the number of previous frames to be outputted at motion detection. Recommended range: 0 to 5, default=0. Do not use large values! Large values will cause Motion to skip video frames and cause unsmooth mpegs. To smooth mpegs use larger values of post_capture instead.

Motion buffers the number of picture frames defined by 'pre_capture'. When motion is detected the pictures in the buffer are included in the video clip generated by ffmpeg. The effect is that it seems the program knew in advance that the event was going to take place and started the recording before it actually happened. This is a nice feature that give more complete video clips of an event.

If pre_capture is set to 0 the feature is disabled. Keep this value below 5.

The recommended value would be approx 0.5 second of video so the value should be defined so it fits the framerate and the desired pre-capture time. E.g. 0.5 second at 20 frames pr second would mean a value of 5. You should never use a value larger than 10.

You can in theory have up to 100s of pre-captured frames but naturally this makes motion leave a larger footprint in the memory of the computer. More important Motion is processing all the buffered images including saving jpegs, encoding mpegs, writing to databases and executing external programs after the first image is detected as Motion.

Motion will not grab another image until this is done. This means that even moderate values for pre_capture combined with high framerates will mean that you will miss quite many frames of Motion. It is therefore recommended to use relative small values for pre_capture. Depending on your chosen framerate and depending on the features enabled values from 1-5 are sensible.

If you wish to create smooth mpegs during events using large pre_capture values will do the opposite! It will create a long pause where a lot of action is missed.

To get a smooth mpeg use a large value for post_capture which does not cost any performance hit or RAM space.

smart_mask_speed

• Type: Integer
• Range / Valid values: 0 - 10
• Default: 0 (disabled)
• Option Topic

Slugginess of the smart mask. Default is 0 = DISABLED. 1 is slow, 10 is fast.

Smartmask is a dynamic, self-learning mask. Smartmask will disable sensitivity in areas with frequent motion (like trees in the wind). Sensitivity is turned on again after some time of no more motion in this area. The built mask is a bit larger at the borders than the actual motion was. This way smartmask works more reliable when sudden moves occur under windy conditions.

smart_mask_speed - tunes the slugginess of the mask. It accepts values from 0 (turned off) to 10 (fast). Fast means here that the mask is built quick, but it is also not staying very long with no more motion. Slow means that it takes a while until the mask is built but it also stays longer. A good start value for smart_mask_speed is 5. This setting is independent from the framerate. The attack and decay time is constant over all available framerates.

When smartmask is enabled and motion is also configured to either write motion-images or motion-mpegs, the current smartmask is copied as an overlay into the black/white motion-pictures/mpegs in red colour. Same thing happens to the webcam stream when Motion runs in setup_mode. That way you can easily adjust smart_mask_speed.

Detailed Description

The mask_file option provides a static mask to turn off sensitivity in certain areas. This is very usefull to mask a street with cars passing by all day long etc...

But imagine a scenario with large bushes and big trees where all the leaves are moving in the wind also triggering motion from time to time even with despeckle turned on. Of course you can also define a static mask here, but what if the bushes are growing during spring and summer? Well, you have to adapt the mask from time to time. What if the camera position moves slightly? What if someone grows new plants in your garden? You always have to setup a new static mask.

The answer to this problem is the smart mask feature introduced in Motion 3.1.18. A dynamic, self-learing mask.

Smart mask will disable sensitivity in areas with frequent motion (like trees in the wind). Sensitivity is turned on again after some time of no more motion in this area. The built mask is a bit larger at the borders than the actual motion. This way smartmask works more reliably when sudden moves occur under windy conditions.

threshold

• Type: Integer
• Range / Valid values: 1 - 2147483647
• Default: 1500
• Option Topic

Threshold for declaring motion. The threshold is the number of changed pixels counted after noise filtering, masking, despeckle, and labelling.

The 'threshold' option is the most important detection setting. When motion runs it compares the current image frame with the previous and counts the number of changed pixels after having processed the image with noise filtering, masking, despeckle and labeling. If more pixels than defined by 'threshold' have changed we assume that we have detected motion. Set the threshold as low as possible so that you get the motion you want detected but large enough so that you do not get detections from noise and plants moving. Note that the larger your frames are, the more pixels you have. So for large picture frame sizes you need a higher threshold.

Use the -s (setup mode) command line option and/or the text_changes config file option to experiment to find the right threshold value. If you do not get small movements detected (see the mouse on the kitchen floor) lower the value. If motion detects too many birds or moving trees, increase the number. Practical values would be from a few hundred to 2000 indoors and 1000-10000 outdoors.

threshold_tune

• Type: Boolean
• Range / Valid values: on, off
• Default: off
• Option Topic

Activates the automatic tuning of threshold level. ( It's broken )

This feature makes Motion continuously adjust the threshold for declaring motion.

The threshold setting is ignored when activating this feature. It may give different results depending on your camera, light conditions, indoor/outdoor, the motion to be detected etc. If it does not work well, deactivate the 'threshold_tune' option and use the manual setting of threshold instead.

-- KennethLavrsen - 05 Apr 2005