Motion Detection Device Using Statistical Error Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion detection systems require complex hardware and high power consumption to detect motion and transmit image sequences over low-power wide area networks, which is inefficient due to the need for processing and data transmission limitations.
Innovation Solution
A method that captures image sequences and detects motion by calculating two-parameter statistical errors between frames, extracting only the moving parts, and transmitting these parts over a network, reducing data transmission and power consumption by using a single device for both motion detection and image capture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a simple motion sensor is used to detect motion, then motion detection capability is achieved, but additional complex camera hardware is required to capture and transmit image sequences
Solution Approach 1:
The patent makes the image captor serve dual functions: both capturing images and detecting motion through statistical error calculation between frames. This eliminates the need for separate motion sensors and reduces hardware complexity while maintaining reliable motion detection capability.
2Device complexity
If a single device is used for both motion detection and image capture, then hardware complexity is reduced, but power consumption increases beyond what the device can supply
Solution Approach 1:
The patent segments the image processing task by dividing frames into arrays of windows and processing only selected windows with high statistical error values. This selective processing significantly reduces computational power requirements and energy consumption while maintaining motion detection accuracy.
Solution Approach 2:
Instead of processing entire frames, the patent applies partial action by calculating statistical errors only for selected windows that are likely to contain motion. This reduces the overall computational load and power consumption to levels suitable for battery-powered or energy-harvesting devices.
3Loss of information
If the entire image sequence is transmitted over the network, then complete image data is available at the CMS, but data transmission exceeds the narrow bandwidth capacity of LPWANs
Solution Approach 1:
The patent extracts only the statistically significant windows from the image sequence that contain motion information. By transmitting only these extracted windows rather than complete frames, the data transmission volume is dramatically reduced to fit within LPWAN bandwidth constraints while preserving essential motion information.
Solution Approach 2:
The patent segments the image data into discrete windows and transmits only those segments containing motion. This segmentation approach allows selective transmission of minimal necessary data, reducing overall transmission volume while maintaining image data completeness for motion-related information.
4Measurement precision
If statistical error calculation is performed on entire frames, then accurate motion detection is achieved, but processing power requirements exceed device capabilities
Solution Approach 1:
The patent divides frames into arrays of windows and performs statistical error calculation only on selected windows rather than entire frames. This segmentation maintains motion detection precision for relevant areas while reducing overall processing power requirements to match device capabilities.
Solution Approach 2:
The patent applies local quality by focusing computational resources on specific windows that are more likely to contain motion based on statistical error thresholds. This localized processing maintains high measurement precision where needed while reducing total processing power consumption.
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
The present invention provides a method of detecting motion of a moving object and transmitting an image of the moving object. The method at least comprises capturing (1) an image sequence of the moving object comprising a plurality of frames separated from each other by a predetermined period of time, taking first (k) and second (k + 1) ones of these frames and detecting whether there is motion in the captured image sequence by calculating (3) a first two-parameter statistical error between the two frames, and determining (4) whether the first two-parameter statistical error is greater than a first predetermined threshold value (T1). If motion is detected, the method also comprises extracting a moving part from the captured image sequence by dividing (6) each of the two frames into corresponding arrays of windows and calculating a second two-parameter statistical error between each of the windows in the first frame (k) and a corresponding one of the windows in the second frame (k + 1), selecting (7, 8) a subset of the windows having the highest value or values of the second two-parameter statistical error between the two frames as the moving part, and transmitting (11) the selected windows over a network. If, on the other hand, motion is not detected, the method instead comprises repeating the same process on the next pair of frames in the captured image sequence. Such a method saves both on the amount of processing and electrical power used before transmission of the windows containing the moving part and on the quantity of data transmitted and electrical power used during transmission. Either or both of the first and second two-parameter statistical errors may, for example, be the mean absolute error (MAE) or the mean squared error (MSE) between corresponding parts of the two frames (k, k + 1). Amongst other things, the invention also provides a motion sensing device for carrying out such a method, wherein the device at least comprises an image captor which can be controlled by a processor of the device both to detect motion of a moving object and to capture an image of the moving object, as well as a wireless transmitter for transmitting the captured image over a network.