DE4224750A1 - Quantitative movement analysis of fish - removing moving objects from video image to leave background, comparing successive images with background and recognising object by difference in intensity - Google Patents

Abstract

Objects are observed with an optical, magnetic, electromagnetic or acoustic sensor. The observations are evaluated with a computer. A continuous or discontinuous reference background picture is received by communication of the individual picture data, by means of which the moved objects are eliminated. During the subsequent observation, the continuous pictures are compared with this background picture. Where the actual picture differs in its intensity from the reference background picture, objects are recognised. A selection may take place of specific object types by criteria such as the object size, the contrast or the colour. The expansion and position of the objects in the actual picture are characterised in a grid as coordinates which are either continually stored or directly processed. From the calculated movement paths, the behaviour of the individual objects are numerically characterised and analysed. USE/ADVANTAGE - For quantitative movement analysis of one or several objects, such as fish, in same measurement space. Enables movements of groups of fish to be analysed before and during feeding.

Description

An optical, magnetic, electromagnetic or acoustic sensor directed. It is connected to a computer that is used for Is capable of image processing. Before the actual measurements becomes a continuous or discontinuous Reference background image by averaging the Single image data recorded, the moving objects be eliminated.

During the measurements themselves, one after each time interval to be specified the current image with the Reference background image compared. The minimum possible Time interval is determined by the frequency of the sensor, e.g. B. one Video camera, limited. However, the time interval can be longer can be selected if the objects to be observed only move slowly. Wherever the current picture of the Reference background image differs, it is assumed that they are objects to be recognized. A selection certain object types by criteria such as object size, the contrast or color is possible. The data that the Extent and position of the existing ones in the current image Characterize objects in a grid to be defined, are either saved continuously or directly processed. The continuous assignment of each Position coordinates representing the object center to the Objects that move around in the same measuring room can be done based on plausibility criteria.

Because of the fast sequence of images z. B. a standard video camera for example 20 ms per field can be assumed that the one coordinate pair of the first image in each second image the closest coordinate pair of the same object listened. The vector that connects these two points characterizes the current direction of movement of the Object and the associated movement speed. they form together with the positions of the objects in the second image the basis for those to be made in the third picture Object assignments. Plausibility checks for the assignment can continue to for the object type to be examined maximum speed from frame to frame and his ability to move continuously or take breaks to make the direction of movement faster or slower change to temporarily disappear behind other objects, Take centrifugal forces, inertia, etc. into account. Such and further information can also be obtained from the trajectories covered in the assignment of the continuous coordinates to the objects are included (self-optimizing tracking system).  

The reconstructed trajectories can be checked can be represented graphically. Through the complete Tracking gives you knowledge of where you are, the direction and speed of movement of each object any time. From the calculated trajectories that becomes Behavior of the objects numerically characterized and analyzed. The characteristics for the behavior criteria can can also be obtained directly during tracking.

Example of an application

The behavior of a group of 6 was monitored with a video camera Fish recorded before and during feeding and after principle described above.

Figs. 1 and 2 show the reconstructed traces of the fish during a 10-second measurement, with the distances traveled by the fish shown in different colors. The associated analysis results are listed in Figs. 3 and 4.

Fig. 1 Reconstructed trajectories at the beginning of feeding. Two animals have already discovered the food falling on the water surface and are excitedly swimming in wild loops.

Fig. 2 Reconstructed trajectories after feeding. The animals swim in tight loops quickly, while the food flakes sink. The latter were eliminated as objects due to their small size.

Fig. 3 Numerical output of the data at the beginning of feeding.

Fig. 4 Numerical output of the data after feeding.

Explanation of the abbreviations in Fig. 3 and 4:
Size = file size, sum of the stored position coordinates
Total = duration of the recording
Sample = duration of the single recording (if total is split)
Interval = frame change speed
No = code letter for the corresponding object
SH = Average location of a single object in the vertical direction in a coarser grid
SHD = standard deviation
PPR = presence of the object during the measurement
TURN = number of turns during the recording
MOTILITY = distance traveled per second in the area commission of the route
MOTDEV = standard deviation for this
DIAGRAM OF MOTILITY = graphic output of motility.
HORIZONTAL POS. = Average location in the horizontal direction, averaged over all objects.
SWIMMING HEIGHT = average location in the vertical direction, averaged over all objects
TURNINGS = number of changes in direction in the x-axis per second, averaged over all objects.

The order of the objects was based on the size of Motility orderly.

Claims (7)

1. Method for quantitative movement analysis of one to many objects in the same measuring room, the objects being observed with an optical, magnetic, electromagnetic or acoustic sensor and the observations being evaluated with a computer, and the following process steps being carried out:

• a) a continuous or discontinuous reference background image is recorded by averaging the individual image data, as a result of which the moving objects are eliminated.
• b) During the subsequent observation, the continuous images are compared with this reference background image.
• c) Objects are recognized where the intensity of the current image differs from the reference background image.
• d) If necessary, certain object types are selected using criteria such as object size, contrast or color.
• e) The extent and position of the objects present in the current image are characterized as coordinates in a grid to be defined.
• f) These coordinates are either saved continuously or processed directly.
• g) The continuous assignment of the position coordinates to the objects that can move around in the same measuring space is carried out according to plausibility criteria for the possible movements of the objects to be observed.
• h) The behavior of the individual objects is numerically characterized and analyzed from the movement paths calculated in this way.

2. The method according to claim 1, characterized in that Local preferences can be determined. 3. The method according to any one of claims 1 to 2, characterized characterized that specific movements are selective be recognized. 4. The method according to any one of claims 1 to 3, characterized characterized that determines the total number of objects becomes. 5. The method according to any one of claims 1 to 4, characterized characterized that for all or individual objects changes over time of movements are documented and be analyzed. 6. The method according to any one of claims 1 to 5, characterized characterized in that the biological effects of noxae living organisms is determined.   7. The method according to any one of claims 1 to 6, characterized characterized that about the behavior of organisms Waters are monitored for noxa.