CN-121982030-A - Urban landscape evaluation system based on image element identification

CN121982030ACN 121982030 ACN121982030 ACN 121982030ACN-121982030-A

Abstract

The invention discloses an urban landscape evaluation system based on image element identification, which relates to the technical field of digital image processing and comprises the following steps of performing color space conversion on a street view image to obtain brightness and chromaticity components; the method comprises the steps of calculating the line-direction gradient energy of a brightness component to locate a ground area, extracting a specular reflection fragment mask based on brightness and chromaticity thresholds, calculating a ground specular reflection fragment index, carrying out depolarization correction on the initial ground boundary density by using the index, and generating a cleanliness evaluation value by combining the non-ground area boundary density. The method can inhibit water film reflection interference and improve accuracy and stability of urban landscape evaluation under complex scenes such as wet road surfaces.

Inventors

CHENG XIAO
ZHANG HUAN
SHI JIEYING
MAO YANYUN

Assignees

浙江大学建筑设计研究院有限公司
浙江大学

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (8)

1. An urban landscape evaluation system based on image element recognition, comprising: The image conversion module is used for acquiring a street view image, carrying out color space conversion on the street view image, and decomposing the street view image to obtain a brightness component, a first chromaticity component and a second chromaticity component; the ground positioning module is used for calculating the line direction gradient energy of the brightness component and determining a ground area mask in the street view image based on the abrupt change characteristic of the energy distribution; a patch detection module for determining a specular reflection patch mask based on the luminance component, the first chrominance component, and the second chrominance component within the ground area mask; The density correction module is used for calculating a ground specular reflection fragmentation index based on the specular reflection fragmentation mask, and correcting the initial boundary density of the ground area of the street view image according to the ground specular reflection fragmentation index to obtain a depolarized ground boundary density; And the grading generation module is used for generating an urban landscape cleanliness evaluation value according to the depolarized ground boundary density and the boundary density of the non-ground area of the street view image.
2. The system for evaluating the urban landscape based on the image element recognition according to claim 1, wherein the step of performing the color space conversion on the street view image to obtain the luminance component, the first chrominance component and the second chrominance component by decomposition comprises the steps of: respectively calculating the average value of all channels of the street view image RGB and the total average value of three channels; carrying out weighted equalization processing on pixel values of channels corresponding to the street view image according to the ratio of the total average value to the average value of each channel to obtain an equalized image; the equalized image is converted from the RGB color space to the CIE Lab color space, the L channel is extracted as a luminance component, the a channel is extracted as a first chrominance component, and the b channel is extracted as a second chrominance component.
3. The image element recognition-based urban landscape evaluation system according to claim 1, wherein calculating the line-direction gradient energy of the luminance component and determining the ground area mask in the street view image based on the abrupt features of the energy distribution comprises: for each line of the image, the gradient magnitudes of all pixels of the line on the luminance component are accumulated to obtain a line energy curve.
4. The image element recognition-based urban landscape evaluation system according to claim 3, wherein determining a ground area mask in the street view image based on abrupt features of the energy distribution comprises: Calculating energy difference values of adjacent rows in the row energy curve, and selecting a row corresponding to the position with the largest difference value as a ground demarcation row; And marking the ground demarcation line and all pixel areas below the ground demarcation line as a ground area mask, and counting the total number of pixels in the area as the ground area.
5. The image element identification-based urban landscape evaluation system according to claim 1, wherein determining a specular reflection patch mask based on the luminance component, the first chrominance component, and the second chrominance component within the ground area mask comprises: Counting brightness component values of all pixels in the ground area mask, and selecting a ninety-fifth percentile value as a brightness threshold; calculating the chromaticity amplitude values of all pixels in the ground area mask, and selecting the number of bits as a chromaticity threshold value, wherein the chromaticity amplitude value is the square root of the sum of squares of the values of the first chromaticity component and the second chromaticity component of the pixels; Traversing each pixel in the ground area mask, judging whether the current pixel simultaneously meets the brightness component which is larger than or equal to the brightness threshold value and the chromaticity amplitude value is smaller than or equal to the chromaticity threshold value, and if so, marking the current pixel as a specular reflection fragment pixel; the set of all the specular patch pixels is formed into a specular patch mask.
6. The image element recognition-based urban landscape evaluation system of claim 4, wherein calculating a ground specular fragmentation index based on the specular fragmentation mask comprises: Marking connected domains on the specular reflection chip mask, and counting the number of the connected domains; Counting the total number of pixels contained in the specular reflection fragment mask to obtain a total reflection area; and calculating the product value of the number of the connected domains and the area of the ground area, and obtaining the ground specular reflection fragmentation index according to the ratio of the product value to the total reflection area.
7. The image element identification-based urban landscape evaluation system according to claim 6, wherein correcting the initial boundary density of the street view image ground area according to the ground specular reflection fragmentation index to obtain the depolarized ground boundary density comprises: Calculating Euclidean distance between a street view image pixel and an adjacent pixel thereof under the CIE Lab color space as a color difference value; determining a color difference threshold value based on an Ojin threshold segmentation method, and marking pixels with color difference values larger than the color difference threshold value as boundary pixels; Counting the number of boundary pixels in the ground area mask, and calculating the ratio of the number of the boundary pixels to the area of the ground area to obtain initial boundary density; and obtaining the depolarized ground boundary density according to the ratio of the initial boundary density to the ground specular reflection fragmentation index.
8. The image element recognition-based urban landscape evaluation system according to claim 7, wherein generating an urban landscape cleanliness evaluation value from the depolarized ground boundary density and the boundary density of the street view image non-ground area comprises: Counting the ratio of the number of boundary pixels of the non-ground area to the area of the non-ground area in the street view image to obtain the non-ground boundary density; obtaining a weight coefficient according to the ratio of the area of the ground area to the number of the full images; weighting and summing the depolarized ground boundary density and the non-ground boundary density according to the weight coefficient to obtain the comprehensive crushing degree; and mapping the comprehensive crushing degree to a preset fraction interval to obtain the urban landscape cleanliness evaluation value.

Description

Urban landscape evaluation system based on image element identification Technical Field The invention relates to the technical field of digital image processing, in particular to an urban landscape evaluation system based on image element identification. Background The urban road and street view cleanliness evaluation is widely applied to environmental sanitation operation assessment, urban fine management, commercial street quality evaluation, travel route optimization and other scenes, and the common practice is to quantify the texture boundary density of road ground and non-ground areas based on street view images, and then output a cleanliness score according to the texture boundary density, so that quick, low-cost and repeatable objective evaluation is realized. The street view image which is actually collected often contains elements such as building facades, greening, signboards, road floors and the like at the same time, is obviously influenced by the gain of shooting equipment, illumination change, shadow shielding and reflection of wet floors after rain, and the high-brightness reflection plaque caused by point light sources such as street lamp lamps easily appears in the floor area, so that the boundary statistics result deviates from the actual structural boundary distribution, further the cleanliness score fluctuation of the same road section under different time and weather conditions is larger, and the requirements of management departments on evaluation stability and comparability are difficult to meet. In the prior art, one type of scheme relies on manual inspection or sampling scoring, has the problems of strong subjectivity, low coverage efficiency, poor consistency across personnel, and the other type of scheme adopts image segmentation or edge detection and other treatments to count image boundaries, but generally only carries out simple color or brightness treatment on the images and then directly carries out threshold segmentation, lacks the equalization treatment steps aiming at inconsistent channel gains and color cast, lacks a reliable positioning and light-reflecting plaque interference suppression mechanism on ground areas, leads to that the light-reflecting plaque can be mistakenly regarded as a large number of fine-broken boundaries to obviously raise the ground boundary density under the conditions of wet road surfaces, strong backlight, night light and the like, finally leads to distortion of comprehensive breakage degree and neatness scores, and is difficult to keep consistency of evaluation results under different acquisition conditions. Disclosure of Invention The invention aims to solve the problems of poor consistency and distortion of urban landscape cleanliness evaluation results caused by interference of ground reflection patches under complex conditions such as wet road surfaces, strong backlight or night lights in the prior art, and provides an urban landscape evaluation system based on image element identification. In order to solve the problems existing in the prior art, the invention adopts the following technical scheme: An urban landscape evaluation system based on image element recognition, comprising: The image conversion module is used for acquiring a street view image, carrying out color space conversion on the street view image, and decomposing the street view image to obtain a brightness component, a first chromaticity component and a second chromaticity component; the ground positioning module is used for calculating the line direction gradient energy of the brightness component and determining a ground area mask in the street view image based on the abrupt change characteristic of the energy distribution; a patch detection module for determining a specular reflection patch mask based on the luminance component, the first chrominance component, and the second chrominance component within the ground area mask; The density correction module is used for calculating a ground specular reflection fragmentation index based on the specular reflection fragmentation mask, and correcting the initial boundary density of the ground area of the street view image according to the ground specular reflection fragmentation index to obtain a depolarized ground boundary density; And the grading generation module is used for generating an urban landscape cleanliness evaluation value according to the depolarized ground boundary density and the boundary density of the non-ground area of the street view image. Preferably, the color space conversion is performed on the street view image, and the decomposition is performed to obtain a luminance component, a first chrominance component and a second chrominance component, including: respectively calculating the average value of all channels of the street view image RGB and the total average value of three channels; carrying out weighted equalization processing on pixel values of channels corresponding to the street view image accor